Last updated: July 22, 2025
Median change data quality checks, SQL examples
This check detects that the median of numeric values has changed more than max_percent from the last measured median.
The median change data quality check has the following variants for each type of data quality checks supported by DQOps.
profile median change
Check description
Verifies that the median in a column changed in a fixed rate since the last readout.
Data quality check name | Friendly name | Category | Check type | Time scale | Quality dimension | Sensor definition | Quality rule | Standard |
---|---|---|---|---|---|---|---|---|
profile_median_change |
Maximum relative change in the median of numeric values since the last known value | anomaly | profiling | Consistency | percentile | change_percent |
Command-line examples
Please expand the section below to see the DQOps command-line examples to run or activate the profile median change data quality check.
Managing profile median change check from DQOps shell
Activate this data quality using the check activate CLI command, providing the connection name, table name, check name, and all other filters. Activates the warning rule with the default parameters.
dqo> check activate -c=connection_name -t=schema_name.table_name -col=column_name -ch=profile_median_change --enable-warning
You can also use patterns to activate the check on all matching tables and columns.
dqo> check activate -c=connection_name -t=schema_prefix*.fact_* -col=column_name -ch=profile_median_change --enable-warning
Additional rule parameters are passed using the -Wrule_parameter_name=value.
Activate this data quality using the check activate CLI command, providing the connection name, table name, check name, and all other filters. Activates the error rule with the default parameters.
dqo> check activate -c=connection_name -t=schema_name.table_name -col=column_name -ch=profile_median_change --enable-error
You can also use patterns to activate the check on all matching tables and columns.
dqo> check activate -c=connection_name -t=schema_prefix*.fact_* -col=column_name -ch=profile_median_change --enable-error
Additional rule parameters are passed using the -Erule_parameter_name=value.
Run this data quality check using the check run CLI command by providing the check name and all other targeting filters. The following example shows how to run the profile_median_change check on all tables and columns on a single data source.
It is also possible to run this check on a specific connection and table. In order to do this, use the connection name and the full table name parameters.
You can also run this check on all tables (and columns) on which the profile_median_change check is enabled using patterns to find tables.
YAML configuration
The sample schema_name.table_name.dqotable.yaml file with the check configured is shown below.
# yaml-language-server: $schema=https://cloud.dqops.com/dqo-yaml-schema/TableYaml-schema.json
apiVersion: dqo/v1
kind: table
spec:
columns:
target_column:
profiling_checks:
anomaly:
profile_median_change:
parameters:
percentile_value: 0.5
warning:
max_percent: 10.0
error:
max_percent: 20.0
fatal:
max_percent: 50.0
labels:
- This is the column that is analyzed for data quality issues
Samples of generated SQL queries for each data source type
Please expand the database engine name section to see the SQL query rendered by a Jinja2 template for the percentile data quality sensor.
BigQuery
{% import '/dialects/bigquery.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }}TIMESTAMP({{ lib.render_time_dimension_expression(table_alias_prefix) }})
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table.`time_period` AS time_period,
nested_table.`time_period_utc` AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE_CONT(
({{ lib.render_target_column('analyzed_table')}}),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
ClickHouse
{% import '/dialects/clickhouse.sql.jinja2' as lib with context -%}
SELECT
quantile({{ parameters.percentile_value }})({{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections_reference('analyzed_table') }}
{{- lib.render_time_dimension_projection_reference('analyzed_table') }}
FROM(
SELECT
original_table.*
{{- lib.render_data_grouping_projections('original_table') }}
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
ORDER BY {{ lib.render_target_column('original_table')}}
) analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
Databricks
{% import '/dialects/databricks.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }}TIMESTAMP({{ lib.render_time_dimension_expression(table_alias_prefix) }})
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table.`time_period` AS time_period,
nested_table.`time_period_utc` AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE(
({{ lib.render_target_column('analyzed_table')}}),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
DB2
{% import '/dialects/db2.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections_reference('analyzed_table') }}
{{- lib.render_time_dimension_projection_reference('analyzed_table') }}
FROM(
SELECT
original_table.*
{{- lib.render_data_grouping_projections('original_table') }}
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
) analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
DuckDB
{% import '/dialects/duckdb.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
HANA
{% import '/dialects/hana.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }} {{ lib.render_time_dimension_expression(table_alias_prefix) }}
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(analyzed_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
time_period,
time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('original_table')}})
OVER (
{%- if lib.data_groupings is not none or lib.time_series is not none %}
PARTITION BY
{%- endif -%}
{{ render_local_time_dimension_projection('original_table') -}}
{{ render_local_data_grouping_projections('original_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
{{- lib.render_where_clause(indentation = ' ', table_alias_prefix='original_table') -}}
) analyzed_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
Oracle
{% import '/dialects/oracle.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections_reference('analyzed_table') }}
{{- lib.render_time_dimension_projection_reference('analyzed_table') }}
FROM(
SELECT
original_table.*
{{- lib.render_data_grouping_projections('original_table') }}
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
) analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
PostgreSQL
{% import '/dialects/postgresql.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
Presto
{% import '/dialects/presto.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }} {{ lib.render_time_dimension_expression(table_alias_prefix) }}
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table."time_period" AS time_period,
nested_table."time_period_utc" AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
APPROX_PERCENTILE(
CAST({{ lib.render_target_column('analyzed_table')}} AS DOUBLE),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
QuestDB
{% import '/dialects/questdb.sql.jinja2' as lib with context -%}
SELECT
APPROX_PERCENTILE({{ lib.render_target_column('analyzed_table')}} * 1.0, {{ parameters.percentile_value }}, 2) AS actual_value
{{- lib.render_data_grouping_projections_reference('analyzed_table') }}
{{- lib.render_time_dimension_projection_reference('analyzed_table') }}
FROM(
SELECT
original_table.*
{{- lib.render_data_grouping_projections('original_table') }}
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
) analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
Redshift
{% import '/dialects/redshift.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
Snowflake
{% import '/dialects/snowflake.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
Spark
{% import '/dialects/spark.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }}TIMESTAMP({{ lib.render_time_dimension_expression(table_alias_prefix) }})
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table.`time_period` AS time_period,
nested_table.`time_period_utc` AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE(
({{ lib.render_target_column('analyzed_table')}}),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SQL Server
{% import '/dialects/sqlserver.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }}CAST({{ lib.render_time_dimension_expression(table_alias_prefix) }} AS DATETIME)
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
{%- macro render_time_period_columns() -%}
{% if lib.time_series is not none -%}
nested_table.[time_period], nested_table.[time_period_utc]
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table.[time_period] AS time_period,
nested_table.[time_period_utc] AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{% if lib.time_series is not none or (data_groupings is not none and (data_groupings | length()) > 0) -%}
GROUP BY {{render_time_period_columns()}} {{- lib.render_data_grouping_projections('analyzed_table', set_leading_comma=(lib.time_series is not none)) }}
ORDER BY {{render_time_period_columns()}} {{- lib.render_data_grouping_projections('analyzed_table', set_leading_comma=(lib.time_series is not none)) }}
{%- endif -%}
Teradata
{% import '/dialects/teradata.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}} * 1.0) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
Trino
{% import '/dialects/trino.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }} {{ lib.render_time_dimension_expression(table_alias_prefix) }}
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table."time_period" AS time_period,
nested_table."time_period_utc" AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
APPROX_PERCENTILE(
CAST({{ lib.render_target_column('analyzed_table')}} AS DOUBLE),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
Expand the Configure with data grouping section to see additional examples for configuring this data quality checks to use data grouping (GROUP BY).
Configuration with data grouping
Sample configuration with data grouping enabled (YAML) The sample below shows how to configure the data grouping and how it affects the generated SQL query.
# yaml-language-server: $schema=https://cloud.dqops.com/dqo-yaml-schema/TableYaml-schema.json
apiVersion: dqo/v1
kind: table
spec:
default_grouping_name: group_by_country_and_state
groupings:
group_by_country_and_state:
level_1:
source: column_value
column: country
level_2:
source: column_value
column: state
columns:
target_column:
profiling_checks:
anomaly:
profile_median_change:
parameters:
percentile_value: 0.5
warning:
max_percent: 10.0
error:
max_percent: 20.0
fatal:
max_percent: 50.0
labels:
- This is the column that is analyzed for data quality issues
country:
labels:
- column used as the first grouping key
state:
labels:
- column used as the second grouping key
Please expand the database engine name section to see the SQL query rendered by a Jinja2 template for the percentile sensor.
BigQuery
{% import '/dialects/bigquery.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }}TIMESTAMP({{ lib.render_time_dimension_expression(table_alias_prefix) }})
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table.`time_period` AS time_period,
nested_table.`time_period_utc` AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE_CONT(
({{ lib.render_target_column('analyzed_table')}}),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
MAX(nested_table.actual_value) AS actual_value,
analyzed_table.`country` AS grouping_level_1,
analyzed_table.`state` AS grouping_level_2
FROM(
SELECT
PERCENTILE_CONT(
(analyzed_table.`target_column`),
0.5)
OVER (PARTITION BY
analyzed_table.`country`,
analyzed_table.`state`
) AS actual_value
FROM `your-google-project-id`.`<target_schema>`.`<target_table>` AS analyzed_table) AS nested_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
ClickHouse
{% import '/dialects/clickhouse.sql.jinja2' as lib with context -%}
SELECT
quantile({{ parameters.percentile_value }})({{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections_reference('analyzed_table') }}
{{- lib.render_time_dimension_projection_reference('analyzed_table') }}
FROM(
SELECT
original_table.*
{{- lib.render_data_grouping_projections('original_table') }}
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
ORDER BY {{ lib.render_target_column('original_table')}}
) analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
quantile(0.5)(analyzed_table."target_column") AS actual_value,
analyzed_table.grouping_level_1,
analyzed_table.grouping_level_2
FROM(
SELECT
original_table.*,
original_table."country" AS grouping_level_1,
original_table."state" AS grouping_level_2
FROM "<target_schema>"."<target_table>" original_table
ORDER BY original_table."target_column"
) analyzed_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
Databricks
{% import '/dialects/databricks.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }}TIMESTAMP({{ lib.render_time_dimension_expression(table_alias_prefix) }})
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table.`time_period` AS time_period,
nested_table.`time_period_utc` AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE(
({{ lib.render_target_column('analyzed_table')}}),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
MAX(nested_table.actual_value) AS actual_value,
analyzed_table.`country` AS grouping_level_1,
analyzed_table.`state` AS grouping_level_2
FROM(
SELECT
PERCENTILE(
(analyzed_table.`target_column`),
0.5)
OVER (PARTITION BY
analyzed_table.`country`,
analyzed_table.`state`
) AS actual_value
FROM `<target_schema>`.`<target_table>` AS analyzed_table) AS nested_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
DB2
{% import '/dialects/db2.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections_reference('analyzed_table') }}
{{- lib.render_time_dimension_projection_reference('analyzed_table') }}
FROM(
SELECT
original_table.*
{{- lib.render_data_grouping_projections('original_table') }}
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
) analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table."target_column") AS actual_value,
analyzed_table.grouping_level_1,
analyzed_table.grouping_level_2
FROM(
SELECT
original_table.*,
original_table."country" AS grouping_level_1,
original_table."state" AS grouping_level_2
FROM "<target_schema>"."<target_table>" original_table
) analyzed_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
DuckDB
{% import '/dialects/duckdb.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
HANA
{% import '/dialects/hana.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }} {{ lib.render_time_dimension_expression(table_alias_prefix) }}
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(analyzed_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
time_period,
time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('original_table')}})
OVER (
{%- if lib.data_groupings is not none or lib.time_series is not none %}
PARTITION BY
{%- endif -%}
{{ render_local_time_dimension_projection('original_table') -}}
{{ render_local_data_grouping_projections('original_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
{{- lib.render_where_clause(indentation = ' ', table_alias_prefix='original_table') -}}
) analyzed_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
MAX(analyzed_table.actual_value) AS actual_value,
analyzed_table."country" AS grouping_level_1,
analyzed_table."state" AS grouping_level_2
FROM(
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY original_table."target_column")
OVER (
PARTITION BY
original_table."country",
original_table."state"
) AS actual_value
FROM "<target_schema>"."<target_table>" original_table) analyzed_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
Oracle
{% import '/dialects/oracle.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections_reference('analyzed_table') }}
{{- lib.render_time_dimension_projection_reference('analyzed_table') }}
FROM(
SELECT
original_table.*
{{- lib.render_data_grouping_projections('original_table') }}
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
) analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table."target_column") AS actual_value,
analyzed_table.grouping_level_1,
analyzed_table.grouping_level_2
FROM(
SELECT
original_table.*,
original_table."country" AS grouping_level_1,
original_table."state" AS grouping_level_2
FROM "<target_schema>"."<target_table>" original_table
) analyzed_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
PostgreSQL
{% import '/dialects/postgresql.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table."target_column") AS actual_value,
analyzed_table."country" AS grouping_level_1,
analyzed_table."state" AS grouping_level_2
FROM "your_postgresql_database"."<target_schema>"."<target_table>" AS analyzed_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
Presto
{% import '/dialects/presto.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }} {{ lib.render_time_dimension_expression(table_alias_prefix) }}
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table."time_period" AS time_period,
nested_table."time_period_utc" AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
APPROX_PERCENTILE(
CAST({{ lib.render_target_column('analyzed_table')}} AS DOUBLE),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
MAX(nested_table.actual_value) AS actual_value,
analyzed_table."country" AS grouping_level_1,
analyzed_table."state" AS grouping_level_2
FROM(
SELECT
APPROX_PERCENTILE(
CAST(analyzed_table."target_column" AS DOUBLE),
0.5)
OVER (PARTITION BY
analyzed_table."country",
analyzed_table."state"
) AS actual_value
FROM "your_trino_database"."<target_schema>"."<target_table>" AS analyzed_table) AS nested_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
QuestDB
{% import '/dialects/questdb.sql.jinja2' as lib with context -%}
SELECT
APPROX_PERCENTILE({{ lib.render_target_column('analyzed_table')}} * 1.0, {{ parameters.percentile_value }}, 2) AS actual_value
{{- lib.render_data_grouping_projections_reference('analyzed_table') }}
{{- lib.render_time_dimension_projection_reference('analyzed_table') }}
FROM(
SELECT
original_table.*
{{- lib.render_data_grouping_projections('original_table') }}
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
) analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
APPROX_PERCENTILE(analyzed_table."target_column" * 1.0, 0.5, 2) AS actual_value,
analyzed_table.grouping_level_1,
analyzed_table.grouping_level_2
FROM(
SELECT
original_table.*,
original_table."country" AS grouping_level_1,
original_table."state" AS grouping_level_2
FROM "<target_table>" original_table
) analyzed_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
Redshift
{% import '/dialects/redshift.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table."target_column") AS actual_value,
analyzed_table."country" AS grouping_level_1,
analyzed_table."state" AS grouping_level_2
FROM "your_redshift_database"."<target_schema>"."<target_table>" AS analyzed_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
Snowflake
{% import '/dialects/snowflake.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table."target_column") AS actual_value,
analyzed_table."country" AS grouping_level_1,
analyzed_table."state" AS grouping_level_2
FROM "your_snowflake_database"."<target_schema>"."<target_table>" AS analyzed_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
Spark
{% import '/dialects/spark.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }}TIMESTAMP({{ lib.render_time_dimension_expression(table_alias_prefix) }})
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table.`time_period` AS time_period,
nested_table.`time_period_utc` AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE(
({{ lib.render_target_column('analyzed_table')}}),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
MAX(nested_table.actual_value) AS actual_value,
analyzed_table.`country` AS grouping_level_1,
analyzed_table.`state` AS grouping_level_2
FROM(
SELECT
PERCENTILE(
(analyzed_table.`target_column`),
0.5)
OVER (PARTITION BY
analyzed_table.`country`,
analyzed_table.`state`
) AS actual_value
FROM `<target_schema>`.`<target_table>` AS analyzed_table) AS nested_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
SQL Server
{% import '/dialects/sqlserver.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }}CAST({{ lib.render_time_dimension_expression(table_alias_prefix) }} AS DATETIME)
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
{%- macro render_time_period_columns() -%}
{% if lib.time_series is not none -%}
nested_table.[time_period], nested_table.[time_period_utc]
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table.[time_period] AS time_period,
nested_table.[time_period_utc] AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{% if lib.time_series is not none or (data_groupings is not none and (data_groupings | length()) > 0) -%}
GROUP BY {{render_time_period_columns()}} {{- lib.render_data_grouping_projections('analyzed_table', set_leading_comma=(lib.time_series is not none)) }}
ORDER BY {{render_time_period_columns()}} {{- lib.render_data_grouping_projections('analyzed_table', set_leading_comma=(lib.time_series is not none)) }}
{%- endif -%}
SELECT
MAX(nested_table.actual_value) AS actual_value,
analyzed_table.[country] AS grouping_level_1,
analyzed_table.[state] AS grouping_level_2
FROM(
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table.[target_column])
OVER (PARTITION BY
analyzed_table.[country],
analyzed_table.[state]
) AS actual_value
FROM [your_sql_server_database].[<target_schema>].[<target_table>] AS analyzed_table) AS nested_table
Teradata
{% import '/dialects/teradata.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}} * 1.0) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table."target_column" * 1.0) AS actual_value,
analyzed_table."country" AS grouping_level_1,
analyzed_table."state" AS grouping_level_2
FROM "<target_schema>"."<target_table>" analyzed_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
Trino
{% import '/dialects/trino.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }} {{ lib.render_time_dimension_expression(table_alias_prefix) }}
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table."time_period" AS time_period,
nested_table."time_period_utc" AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
APPROX_PERCENTILE(
CAST({{ lib.render_target_column('analyzed_table')}} AS DOUBLE),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
MAX(nested_table.actual_value) AS actual_value,
analyzed_table."country" AS grouping_level_1,
analyzed_table."state" AS grouping_level_2
FROM(
SELECT
APPROX_PERCENTILE(
CAST(analyzed_table."target_column" AS DOUBLE),
0.5)
OVER (PARTITION BY
analyzed_table."country",
analyzed_table."state"
) AS actual_value
FROM "your_trino_catalog"."<target_schema>"."<target_table>" AS analyzed_table) AS nested_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
daily median change
Check description
Verifies that the median in a column changed in a fixed rate since the last readout.
Data quality check name | Friendly name | Category | Check type | Time scale | Quality dimension | Sensor definition | Quality rule | Standard |
---|---|---|---|---|---|---|---|---|
daily_median_change |
Maximum relative change in the median of numeric values since the last known value | anomaly | monitoring | daily | Consistency | percentile | change_percent |
Command-line examples
Please expand the section below to see the DQOps command-line examples to run or activate the daily median change data quality check.
Managing daily median change check from DQOps shell
Activate this data quality using the check activate CLI command, providing the connection name, table name, check name, and all other filters. Activates the warning rule with the default parameters.
dqo> check activate -c=connection_name -t=schema_name.table_name -col=column_name -ch=daily_median_change --enable-warning
You can also use patterns to activate the check on all matching tables and columns.
dqo> check activate -c=connection_name -t=schema_prefix*.fact_* -col=column_name -ch=daily_median_change --enable-warning
Additional rule parameters are passed using the -Wrule_parameter_name=value.
Activate this data quality using the check activate CLI command, providing the connection name, table name, check name, and all other filters. Activates the error rule with the default parameters.
dqo> check activate -c=connection_name -t=schema_name.table_name -col=column_name -ch=daily_median_change --enable-error
You can also use patterns to activate the check on all matching tables and columns.
dqo> check activate -c=connection_name -t=schema_prefix*.fact_* -col=column_name -ch=daily_median_change --enable-error
Additional rule parameters are passed using the -Erule_parameter_name=value.
Run this data quality check using the check run CLI command by providing the check name and all other targeting filters. The following example shows how to run the daily_median_change check on all tables and columns on a single data source.
It is also possible to run this check on a specific connection and table. In order to do this, use the connection name and the full table name parameters.
You can also run this check on all tables (and columns) on which the daily_median_change check is enabled using patterns to find tables.
YAML configuration
The sample schema_name.table_name.dqotable.yaml file with the check configured is shown below.
# yaml-language-server: $schema=https://cloud.dqops.com/dqo-yaml-schema/TableYaml-schema.json
apiVersion: dqo/v1
kind: table
spec:
columns:
target_column:
monitoring_checks:
daily:
anomaly:
daily_median_change:
parameters:
percentile_value: 0.5
warning:
max_percent: 10.0
error:
max_percent: 20.0
fatal:
max_percent: 50.0
labels:
- This is the column that is analyzed for data quality issues
Samples of generated SQL queries for each data source type
Please expand the database engine name section to see the SQL query rendered by a Jinja2 template for the percentile data quality sensor.
BigQuery
{% import '/dialects/bigquery.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }}TIMESTAMP({{ lib.render_time_dimension_expression(table_alias_prefix) }})
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table.`time_period` AS time_period,
nested_table.`time_period_utc` AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE_CONT(
({{ lib.render_target_column('analyzed_table')}}),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
ClickHouse
{% import '/dialects/clickhouse.sql.jinja2' as lib with context -%}
SELECT
quantile({{ parameters.percentile_value }})({{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections_reference('analyzed_table') }}
{{- lib.render_time_dimension_projection_reference('analyzed_table') }}
FROM(
SELECT
original_table.*
{{- lib.render_data_grouping_projections('original_table') }}
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
ORDER BY {{ lib.render_target_column('original_table')}}
) analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
Databricks
{% import '/dialects/databricks.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }}TIMESTAMP({{ lib.render_time_dimension_expression(table_alias_prefix) }})
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table.`time_period` AS time_period,
nested_table.`time_period_utc` AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE(
({{ lib.render_target_column('analyzed_table')}}),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
DB2
{% import '/dialects/db2.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections_reference('analyzed_table') }}
{{- lib.render_time_dimension_projection_reference('analyzed_table') }}
FROM(
SELECT
original_table.*
{{- lib.render_data_grouping_projections('original_table') }}
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
) analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
DuckDB
{% import '/dialects/duckdb.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
HANA
{% import '/dialects/hana.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }} {{ lib.render_time_dimension_expression(table_alias_prefix) }}
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(analyzed_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
time_period,
time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('original_table')}})
OVER (
{%- if lib.data_groupings is not none or lib.time_series is not none %}
PARTITION BY
{%- endif -%}
{{ render_local_time_dimension_projection('original_table') -}}
{{ render_local_data_grouping_projections('original_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
{{- lib.render_where_clause(indentation = ' ', table_alias_prefix='original_table') -}}
) analyzed_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
Oracle
{% import '/dialects/oracle.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections_reference('analyzed_table') }}
{{- lib.render_time_dimension_projection_reference('analyzed_table') }}
FROM(
SELECT
original_table.*
{{- lib.render_data_grouping_projections('original_table') }}
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
) analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
PostgreSQL
{% import '/dialects/postgresql.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
Presto
{% import '/dialects/presto.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }} {{ lib.render_time_dimension_expression(table_alias_prefix) }}
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table."time_period" AS time_period,
nested_table."time_period_utc" AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
APPROX_PERCENTILE(
CAST({{ lib.render_target_column('analyzed_table')}} AS DOUBLE),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
QuestDB
{% import '/dialects/questdb.sql.jinja2' as lib with context -%}
SELECT
APPROX_PERCENTILE({{ lib.render_target_column('analyzed_table')}} * 1.0, {{ parameters.percentile_value }}, 2) AS actual_value
{{- lib.render_data_grouping_projections_reference('analyzed_table') }}
{{- lib.render_time_dimension_projection_reference('analyzed_table') }}
FROM(
SELECT
original_table.*
{{- lib.render_data_grouping_projections('original_table') }}
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
) analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
Redshift
{% import '/dialects/redshift.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
Snowflake
{% import '/dialects/snowflake.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
Spark
{% import '/dialects/spark.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }}TIMESTAMP({{ lib.render_time_dimension_expression(table_alias_prefix) }})
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table.`time_period` AS time_period,
nested_table.`time_period_utc` AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE(
({{ lib.render_target_column('analyzed_table')}}),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SQL Server
{% import '/dialects/sqlserver.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }}CAST({{ lib.render_time_dimension_expression(table_alias_prefix) }} AS DATETIME)
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
{%- macro render_time_period_columns() -%}
{% if lib.time_series is not none -%}
nested_table.[time_period], nested_table.[time_period_utc]
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table.[time_period] AS time_period,
nested_table.[time_period_utc] AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{% if lib.time_series is not none or (data_groupings is not none and (data_groupings | length()) > 0) -%}
GROUP BY {{render_time_period_columns()}} {{- lib.render_data_grouping_projections('analyzed_table', set_leading_comma=(lib.time_series is not none)) }}
ORDER BY {{render_time_period_columns()}} {{- lib.render_data_grouping_projections('analyzed_table', set_leading_comma=(lib.time_series is not none)) }}
{%- endif -%}
Teradata
{% import '/dialects/teradata.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}} * 1.0) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
Trino
{% import '/dialects/trino.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }} {{ lib.render_time_dimension_expression(table_alias_prefix) }}
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table."time_period" AS time_period,
nested_table."time_period_utc" AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
APPROX_PERCENTILE(
CAST({{ lib.render_target_column('analyzed_table')}} AS DOUBLE),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
Expand the Configure with data grouping section to see additional examples for configuring this data quality checks to use data grouping (GROUP BY).
Configuration with data grouping
Sample configuration with data grouping enabled (YAML) The sample below shows how to configure the data grouping and how it affects the generated SQL query.
# yaml-language-server: $schema=https://cloud.dqops.com/dqo-yaml-schema/TableYaml-schema.json
apiVersion: dqo/v1
kind: table
spec:
default_grouping_name: group_by_country_and_state
groupings:
group_by_country_and_state:
level_1:
source: column_value
column: country
level_2:
source: column_value
column: state
columns:
target_column:
monitoring_checks:
daily:
anomaly:
daily_median_change:
parameters:
percentile_value: 0.5
warning:
max_percent: 10.0
error:
max_percent: 20.0
fatal:
max_percent: 50.0
labels:
- This is the column that is analyzed for data quality issues
country:
labels:
- column used as the first grouping key
state:
labels:
- column used as the second grouping key
Please expand the database engine name section to see the SQL query rendered by a Jinja2 template for the percentile sensor.
BigQuery
{% import '/dialects/bigquery.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }}TIMESTAMP({{ lib.render_time_dimension_expression(table_alias_prefix) }})
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table.`time_period` AS time_period,
nested_table.`time_period_utc` AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE_CONT(
({{ lib.render_target_column('analyzed_table')}}),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
MAX(nested_table.actual_value) AS actual_value,
analyzed_table.`country` AS grouping_level_1,
analyzed_table.`state` AS grouping_level_2
FROM(
SELECT
PERCENTILE_CONT(
(analyzed_table.`target_column`),
0.5)
OVER (PARTITION BY
analyzed_table.`country`,
analyzed_table.`state`
) AS actual_value
FROM `your-google-project-id`.`<target_schema>`.`<target_table>` AS analyzed_table) AS nested_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
ClickHouse
{% import '/dialects/clickhouse.sql.jinja2' as lib with context -%}
SELECT
quantile({{ parameters.percentile_value }})({{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections_reference('analyzed_table') }}
{{- lib.render_time_dimension_projection_reference('analyzed_table') }}
FROM(
SELECT
original_table.*
{{- lib.render_data_grouping_projections('original_table') }}
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
ORDER BY {{ lib.render_target_column('original_table')}}
) analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
quantile(0.5)(analyzed_table."target_column") AS actual_value,
analyzed_table.grouping_level_1,
analyzed_table.grouping_level_2
FROM(
SELECT
original_table.*,
original_table."country" AS grouping_level_1,
original_table."state" AS grouping_level_2
FROM "<target_schema>"."<target_table>" original_table
ORDER BY original_table."target_column"
) analyzed_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
Databricks
{% import '/dialects/databricks.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }}TIMESTAMP({{ lib.render_time_dimension_expression(table_alias_prefix) }})
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table.`time_period` AS time_period,
nested_table.`time_period_utc` AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE(
({{ lib.render_target_column('analyzed_table')}}),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
MAX(nested_table.actual_value) AS actual_value,
analyzed_table.`country` AS grouping_level_1,
analyzed_table.`state` AS grouping_level_2
FROM(
SELECT
PERCENTILE(
(analyzed_table.`target_column`),
0.5)
OVER (PARTITION BY
analyzed_table.`country`,
analyzed_table.`state`
) AS actual_value
FROM `<target_schema>`.`<target_table>` AS analyzed_table) AS nested_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
DB2
{% import '/dialects/db2.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections_reference('analyzed_table') }}
{{- lib.render_time_dimension_projection_reference('analyzed_table') }}
FROM(
SELECT
original_table.*
{{- lib.render_data_grouping_projections('original_table') }}
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
) analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table."target_column") AS actual_value,
analyzed_table.grouping_level_1,
analyzed_table.grouping_level_2
FROM(
SELECT
original_table.*,
original_table."country" AS grouping_level_1,
original_table."state" AS grouping_level_2
FROM "<target_schema>"."<target_table>" original_table
) analyzed_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
DuckDB
{% import '/dialects/duckdb.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
HANA
{% import '/dialects/hana.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }} {{ lib.render_time_dimension_expression(table_alias_prefix) }}
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(analyzed_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
time_period,
time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('original_table')}})
OVER (
{%- if lib.data_groupings is not none or lib.time_series is not none %}
PARTITION BY
{%- endif -%}
{{ render_local_time_dimension_projection('original_table') -}}
{{ render_local_data_grouping_projections('original_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
{{- lib.render_where_clause(indentation = ' ', table_alias_prefix='original_table') -}}
) analyzed_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
MAX(analyzed_table.actual_value) AS actual_value,
analyzed_table."country" AS grouping_level_1,
analyzed_table."state" AS grouping_level_2
FROM(
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY original_table."target_column")
OVER (
PARTITION BY
original_table."country",
original_table."state"
) AS actual_value
FROM "<target_schema>"."<target_table>" original_table) analyzed_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
Oracle
{% import '/dialects/oracle.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections_reference('analyzed_table') }}
{{- lib.render_time_dimension_projection_reference('analyzed_table') }}
FROM(
SELECT
original_table.*
{{- lib.render_data_grouping_projections('original_table') }}
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
) analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table."target_column") AS actual_value,
analyzed_table.grouping_level_1,
analyzed_table.grouping_level_2
FROM(
SELECT
original_table.*,
original_table."country" AS grouping_level_1,
original_table."state" AS grouping_level_2
FROM "<target_schema>"."<target_table>" original_table
) analyzed_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
PostgreSQL
{% import '/dialects/postgresql.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table."target_column") AS actual_value,
analyzed_table."country" AS grouping_level_1,
analyzed_table."state" AS grouping_level_2
FROM "your_postgresql_database"."<target_schema>"."<target_table>" AS analyzed_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
Presto
{% import '/dialects/presto.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }} {{ lib.render_time_dimension_expression(table_alias_prefix) }}
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table."time_period" AS time_period,
nested_table."time_period_utc" AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
APPROX_PERCENTILE(
CAST({{ lib.render_target_column('analyzed_table')}} AS DOUBLE),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
MAX(nested_table.actual_value) AS actual_value,
analyzed_table."country" AS grouping_level_1,
analyzed_table."state" AS grouping_level_2
FROM(
SELECT
APPROX_PERCENTILE(
CAST(analyzed_table."target_column" AS DOUBLE),
0.5)
OVER (PARTITION BY
analyzed_table."country",
analyzed_table."state"
) AS actual_value
FROM "your_trino_database"."<target_schema>"."<target_table>" AS analyzed_table) AS nested_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
QuestDB
{% import '/dialects/questdb.sql.jinja2' as lib with context -%}
SELECT
APPROX_PERCENTILE({{ lib.render_target_column('analyzed_table')}} * 1.0, {{ parameters.percentile_value }}, 2) AS actual_value
{{- lib.render_data_grouping_projections_reference('analyzed_table') }}
{{- lib.render_time_dimension_projection_reference('analyzed_table') }}
FROM(
SELECT
original_table.*
{{- lib.render_data_grouping_projections('original_table') }}
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
) analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
APPROX_PERCENTILE(analyzed_table."target_column" * 1.0, 0.5, 2) AS actual_value,
analyzed_table.grouping_level_1,
analyzed_table.grouping_level_2
FROM(
SELECT
original_table.*,
original_table."country" AS grouping_level_1,
original_table."state" AS grouping_level_2
FROM "<target_table>" original_table
) analyzed_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
Redshift
{% import '/dialects/redshift.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table."target_column") AS actual_value,
analyzed_table."country" AS grouping_level_1,
analyzed_table."state" AS grouping_level_2
FROM "your_redshift_database"."<target_schema>"."<target_table>" AS analyzed_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
Snowflake
{% import '/dialects/snowflake.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table."target_column") AS actual_value,
analyzed_table."country" AS grouping_level_1,
analyzed_table."state" AS grouping_level_2
FROM "your_snowflake_database"."<target_schema>"."<target_table>" AS analyzed_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
Spark
{% import '/dialects/spark.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }}TIMESTAMP({{ lib.render_time_dimension_expression(table_alias_prefix) }})
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table.`time_period` AS time_period,
nested_table.`time_period_utc` AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE(
({{ lib.render_target_column('analyzed_table')}}),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
MAX(nested_table.actual_value) AS actual_value,
analyzed_table.`country` AS grouping_level_1,
analyzed_table.`state` AS grouping_level_2
FROM(
SELECT
PERCENTILE(
(analyzed_table.`target_column`),
0.5)
OVER (PARTITION BY
analyzed_table.`country`,
analyzed_table.`state`
) AS actual_value
FROM `<target_schema>`.`<target_table>` AS analyzed_table) AS nested_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
SQL Server
{% import '/dialects/sqlserver.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }}CAST({{ lib.render_time_dimension_expression(table_alias_prefix) }} AS DATETIME)
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
{%- macro render_time_period_columns() -%}
{% if lib.time_series is not none -%}
nested_table.[time_period], nested_table.[time_period_utc]
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table.[time_period] AS time_period,
nested_table.[time_period_utc] AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{% if lib.time_series is not none or (data_groupings is not none and (data_groupings | length()) > 0) -%}
GROUP BY {{render_time_period_columns()}} {{- lib.render_data_grouping_projections('analyzed_table', set_leading_comma=(lib.time_series is not none)) }}
ORDER BY {{render_time_period_columns()}} {{- lib.render_data_grouping_projections('analyzed_table', set_leading_comma=(lib.time_series is not none)) }}
{%- endif -%}
SELECT
MAX(nested_table.actual_value) AS actual_value,
analyzed_table.[country] AS grouping_level_1,
analyzed_table.[state] AS grouping_level_2
FROM(
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table.[target_column])
OVER (PARTITION BY
analyzed_table.[country],
analyzed_table.[state]
) AS actual_value
FROM [your_sql_server_database].[<target_schema>].[<target_table>] AS analyzed_table) AS nested_table
Teradata
{% import '/dialects/teradata.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}} * 1.0) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table."target_column" * 1.0) AS actual_value,
analyzed_table."country" AS grouping_level_1,
analyzed_table."state" AS grouping_level_2
FROM "<target_schema>"."<target_table>" analyzed_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
Trino
{% import '/dialects/trino.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }} {{ lib.render_time_dimension_expression(table_alias_prefix) }}
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table."time_period" AS time_period,
nested_table."time_period_utc" AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
APPROX_PERCENTILE(
CAST({{ lib.render_target_column('analyzed_table')}} AS DOUBLE),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
MAX(nested_table.actual_value) AS actual_value,
analyzed_table."country" AS grouping_level_1,
analyzed_table."state" AS grouping_level_2
FROM(
SELECT
APPROX_PERCENTILE(
CAST(analyzed_table."target_column" AS DOUBLE),
0.5)
OVER (PARTITION BY
analyzed_table."country",
analyzed_table."state"
) AS actual_value
FROM "your_trino_catalog"."<target_schema>"."<target_table>" AS analyzed_table) AS nested_table
GROUP BY grouping_level_1, grouping_level_2
ORDER BY grouping_level_1, grouping_level_2
daily partition median change
Check description
Verifies that the median in a column changed in a fixed rate since the last readout.
Data quality check name | Friendly name | Category | Check type | Time scale | Quality dimension | Sensor definition | Quality rule | Standard |
---|---|---|---|---|---|---|---|---|
daily_partition_median_change |
Maximum relative change in the median of numeric values since the last known value | anomaly | partitioned | daily | Consistency | percentile | change_percent |
Command-line examples
Please expand the section below to see the DQOps command-line examples to run or activate the daily partition median change data quality check.
Managing daily partition median change check from DQOps shell
Activate this data quality using the check activate CLI command, providing the connection name, table name, check name, and all other filters. Activates the warning rule with the default parameters.
dqo> check activate -c=connection_name -t=schema_name.table_name -col=column_name -ch=daily_partition_median_change --enable-warning
You can also use patterns to activate the check on all matching tables and columns.
dqo> check activate -c=connection_name -t=schema_prefix*.fact_* -col=column_name -ch=daily_partition_median_change --enable-warning
Additional rule parameters are passed using the -Wrule_parameter_name=value.
Activate this data quality using the check activate CLI command, providing the connection name, table name, check name, and all other filters. Activates the error rule with the default parameters.
dqo> check activate -c=connection_name -t=schema_name.table_name -col=column_name -ch=daily_partition_median_change --enable-error
You can also use patterns to activate the check on all matching tables and columns.
dqo> check activate -c=connection_name -t=schema_prefix*.fact_* -col=column_name -ch=daily_partition_median_change --enable-error
Additional rule parameters are passed using the -Erule_parameter_name=value.
Run this data quality check using the check run CLI command by providing the check name and all other targeting filters. The following example shows how to run the daily_partition_median_change check on all tables and columns on a single data source.
It is also possible to run this check on a specific connection and table. In order to do this, use the connection name and the full table name parameters.
You can also run this check on all tables (and columns) on which the daily_partition_median_change check is enabled using patterns to find tables.
YAML configuration
The sample schema_name.table_name.dqotable.yaml file with the check configured is shown below.
# yaml-language-server: $schema=https://cloud.dqops.com/dqo-yaml-schema/TableYaml-schema.json
apiVersion: dqo/v1
kind: table
spec:
timestamp_columns:
partition_by_column: date_column
incremental_time_window:
daily_partitioning_recent_days: 7
monthly_partitioning_recent_months: 1
columns:
target_column:
partitioned_checks:
daily:
anomaly:
daily_partition_median_change:
parameters:
percentile_value: 0.5
warning:
max_percent: 10.0
error:
max_percent: 20.0
fatal:
max_percent: 50.0
labels:
- This is the column that is analyzed for data quality issues
date_column:
labels:
- "date or datetime column used as a daily or monthly partitioning key, dates\
\ (and times) are truncated to a day or a month by the sensor's query for\
\ partitioned checks"
Samples of generated SQL queries for each data source type
Please expand the database engine name section to see the SQL query rendered by a Jinja2 template for the percentile data quality sensor.
BigQuery
{% import '/dialects/bigquery.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }}TIMESTAMP({{ lib.render_time_dimension_expression(table_alias_prefix) }})
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table.`time_period` AS time_period,
nested_table.`time_period_utc` AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE_CONT(
({{ lib.render_target_column('analyzed_table')}}),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
MAX(nested_table.actual_value) AS actual_value,
nested_table.`time_period` AS time_period,
nested_table.`time_period_utc` AS time_period_utc
FROM(
SELECT
PERCENTILE_CONT(
(analyzed_table.`target_column`),
0.5)
OVER (PARTITION BY
CAST(analyzed_table.`date_column` AS DATE),
TIMESTAMP(CAST(analyzed_table.`date_column` AS DATE))
) AS actual_value,
CAST(analyzed_table.`date_column` AS DATE) AS time_period,
TIMESTAMP(CAST(analyzed_table.`date_column` AS DATE)) AS time_period_utc
FROM `your-google-project-id`.`<target_schema>`.`<target_table>` AS analyzed_table) AS nested_table
GROUP BY time_period, time_period_utc
ORDER BY time_period, time_period_utc
ClickHouse
{% import '/dialects/clickhouse.sql.jinja2' as lib with context -%}
SELECT
quantile({{ parameters.percentile_value }})({{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections_reference('analyzed_table') }}
{{- lib.render_time_dimension_projection_reference('analyzed_table') }}
FROM(
SELECT
original_table.*
{{- lib.render_data_grouping_projections('original_table') }}
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
ORDER BY {{ lib.render_target_column('original_table')}}
) analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
quantile(0.5)(analyzed_table."target_column") AS actual_value,
time_period,
time_period_utc
FROM(
SELECT
original_table.*,
CAST(original_table."date_column" AS DATE) AS time_period,
toDateTime64(CAST(original_table."date_column" AS DATE), 3) AS time_period_utc
FROM "<target_schema>"."<target_table>" original_table
ORDER BY original_table."target_column"
) analyzed_table
GROUP BY time_period, time_period_utc
ORDER BY time_period, time_period_utc
Databricks
{% import '/dialects/databricks.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }}TIMESTAMP({{ lib.render_time_dimension_expression(table_alias_prefix) }})
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table.`time_period` AS time_period,
nested_table.`time_period_utc` AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE(
({{ lib.render_target_column('analyzed_table')}}),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
MAX(nested_table.actual_value) AS actual_value,
nested_table.`time_period` AS time_period,
nested_table.`time_period_utc` AS time_period_utc
FROM(
SELECT
PERCENTILE(
(analyzed_table.`target_column`),
0.5)
OVER (PARTITION BY
CAST(analyzed_table.`date_column` AS DATE),
TIMESTAMP(CAST(analyzed_table.`date_column` AS DATE))
) AS actual_value,
CAST(analyzed_table.`date_column` AS DATE) AS time_period,
TIMESTAMP(CAST(analyzed_table.`date_column` AS DATE)) AS time_period_utc
FROM `<target_schema>`.`<target_table>` AS analyzed_table) AS nested_table
GROUP BY time_period, time_period_utc
ORDER BY time_period, time_period_utc
DB2
{% import '/dialects/db2.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections_reference('analyzed_table') }}
{{- lib.render_time_dimension_projection_reference('analyzed_table') }}
FROM(
SELECT
original_table.*
{{- lib.render_data_grouping_projections('original_table') }}
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
) analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table."target_column") AS actual_value,
time_period,
time_period_utc
FROM(
SELECT
original_table.*,
CAST(original_table."date_column" AS DATE) AS time_period,
TIMESTAMP(CAST(original_table."date_column" AS DATE)) AS time_period_utc
FROM "<target_schema>"."<target_table>" original_table
) analyzed_table
GROUP BY time_period, time_period_utc
ORDER BY time_period, time_period_utc
DuckDB
{% import '/dialects/duckdb.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table."target_column") AS actual_value,
CAST(analyzed_table."date_column" AS date) AS time_period,
CAST((CAST(analyzed_table."date_column" AS date)) AS TIMESTAMP WITH TIME ZONE) AS time_period_utc
FROM AS analyzed_table
GROUP BY time_period, time_period_utc
ORDER BY time_period, time_period_utc
HANA
{% import '/dialects/hana.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }} {{ lib.render_time_dimension_expression(table_alias_prefix) }}
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(analyzed_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
time_period,
time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('original_table')}})
OVER (
{%- if lib.data_groupings is not none or lib.time_series is not none %}
PARTITION BY
{%- endif -%}
{{ render_local_time_dimension_projection('original_table') -}}
{{ render_local_data_grouping_projections('original_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
{{- lib.render_where_clause(indentation = ' ', table_alias_prefix='original_table') -}}
) analyzed_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
MAX(analyzed_table.actual_value) AS actual_value,
time_period,
time_period_utc
FROM(
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY original_table."target_column")
OVER (
PARTITION BY
CAST(original_table."date_column" AS DATE),
CAST(original_table."date_column" AS DATE)
) AS actual_value,
CAST(original_table."date_column" AS DATE) AS time_period,
TO_TIMESTAMP(CAST(original_table."date_column" AS DATE)) AS time_period_utc
FROM "<target_schema>"."<target_table>" original_table) analyzed_table
GROUP BY time_period, time_period_utc
ORDER BY time_period, time_period_utc
Oracle
{% import '/dialects/oracle.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections_reference('analyzed_table') }}
{{- lib.render_time_dimension_projection_reference('analyzed_table') }}
FROM(
SELECT
original_table.*
{{- lib.render_data_grouping_projections('original_table') }}
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
) analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table."target_column") AS actual_value,
time_period,
time_period_utc
FROM(
SELECT
original_table.*,
TRUNC(CAST(original_table."date_column" AS DATE)) AS time_period,
CAST(TRUNC(CAST(original_table."date_column" AS DATE)) AS TIMESTAMP WITH TIME ZONE) AS time_period_utc
FROM "<target_schema>"."<target_table>" original_table
) analyzed_table
GROUP BY time_period, time_period_utc
ORDER BY time_period, time_period_utc
PostgreSQL
{% import '/dialects/postgresql.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table."target_column") AS actual_value,
CAST(analyzed_table."date_column" AS date) AS time_period,
CAST((CAST(analyzed_table."date_column" AS date)) AS TIMESTAMP WITH TIME ZONE) AS time_period_utc
FROM "your_postgresql_database"."<target_schema>"."<target_table>" AS analyzed_table
GROUP BY time_period, time_period_utc
ORDER BY time_period, time_period_utc
Presto
{% import '/dialects/presto.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }} {{ lib.render_time_dimension_expression(table_alias_prefix) }}
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table."time_period" AS time_period,
nested_table."time_period_utc" AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
APPROX_PERCENTILE(
CAST({{ lib.render_target_column('analyzed_table')}} AS DOUBLE),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
MAX(nested_table.actual_value) AS actual_value,
nested_table."time_period" AS time_period,
nested_table."time_period_utc" AS time_period_utc
FROM(
SELECT
APPROX_PERCENTILE(
CAST(analyzed_table."target_column" AS DOUBLE),
0.5)
OVER (PARTITION BY
CAST(analyzed_table."date_column" AS date),
CAST(analyzed_table."date_column" AS date)
) AS actual_value,
CAST(analyzed_table."date_column" AS date) AS time_period,
CAST(CAST(analyzed_table."date_column" AS date) AS TIMESTAMP) AS time_period_utc
FROM "your_trino_database"."<target_schema>"."<target_table>" AS analyzed_table) AS nested_table
GROUP BY time_period, time_period_utc
ORDER BY time_period, time_period_utc
QuestDB
{% import '/dialects/questdb.sql.jinja2' as lib with context -%}
SELECT
APPROX_PERCENTILE({{ lib.render_target_column('analyzed_table')}} * 1.0, {{ parameters.percentile_value }}, 2) AS actual_value
{{- lib.render_data_grouping_projections_reference('analyzed_table') }}
{{- lib.render_time_dimension_projection_reference('analyzed_table') }}
FROM(
SELECT
original_table.*
{{- lib.render_data_grouping_projections('original_table') }}
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
) analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
APPROX_PERCENTILE(analyzed_table."target_column" * 1.0, 0.5, 2) AS actual_value,
time_period,
time_period_utc
FROM(
SELECT
original_table.*,
CAST(DATE_TRUNC('day', original_table."date_column") AS DATE) AS time_period,
CAST((CAST(DATE_TRUNC('day', original_table."date_column") AS DATE)) AS TIMESTAMP WITH TIME ZONE) AS time_period_utc
FROM "<target_table>" original_table
) analyzed_table
GROUP BY time_period, time_period_utc
ORDER BY time_period, time_period_utc
Redshift
{% import '/dialects/redshift.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table."target_column") AS actual_value,
CAST(analyzed_table."date_column" AS date) AS time_period,
CAST((CAST(analyzed_table."date_column" AS date)) AS TIMESTAMP WITH TIME ZONE) AS time_period_utc
FROM "your_redshift_database"."<target_schema>"."<target_table>" AS analyzed_table
GROUP BY time_period, time_period_utc
ORDER BY time_period, time_period_utc
Snowflake
{% import '/dialects/snowflake.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table."target_column") AS actual_value,
CAST(analyzed_table."date_column" AS date) AS time_period,
TO_TIMESTAMP(CAST(analyzed_table."date_column" AS date)) AS time_period_utc
FROM "your_snowflake_database"."<target_schema>"."<target_table>" AS analyzed_table
GROUP BY time_period, time_period_utc
ORDER BY time_period, time_period_utc
Spark
{% import '/dialects/spark.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }}TIMESTAMP({{ lib.render_time_dimension_expression(table_alias_prefix) }})
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table.`time_period` AS time_period,
nested_table.`time_period_utc` AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE(
({{ lib.render_target_column('analyzed_table')}}),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
MAX(nested_table.actual_value) AS actual_value,
nested_table.`time_period` AS time_period,
nested_table.`time_period_utc` AS time_period_utc
FROM(
SELECT
PERCENTILE(
(analyzed_table.`target_column`),
0.5)
OVER (PARTITION BY
CAST(analyzed_table.`date_column` AS DATE),
TIMESTAMP(CAST(analyzed_table.`date_column` AS DATE))
) AS actual_value,
CAST(analyzed_table.`date_column` AS DATE) AS time_period,
TIMESTAMP(CAST(analyzed_table.`date_column` AS DATE)) AS time_period_utc
FROM `<target_schema>`.`<target_table>` AS analyzed_table) AS nested_table
GROUP BY time_period, time_period_utc
ORDER BY time_period, time_period_utc
SQL Server
{% import '/dialects/sqlserver.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }}CAST({{ lib.render_time_dimension_expression(table_alias_prefix) }} AS DATETIME)
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
{%- macro render_time_period_columns() -%}
{% if lib.time_series is not none -%}
nested_table.[time_period], nested_table.[time_period_utc]
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table.[time_period] AS time_period,
nested_table.[time_period_utc] AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{% if lib.time_series is not none or (data_groupings is not none and (data_groupings | length()) > 0) -%}
GROUP BY {{render_time_period_columns()}} {{- lib.render_data_grouping_projections('analyzed_table', set_leading_comma=(lib.time_series is not none)) }}
ORDER BY {{render_time_period_columns()}} {{- lib.render_data_grouping_projections('analyzed_table', set_leading_comma=(lib.time_series is not none)) }}
{%- endif -%}
SELECT
MAX(nested_table.actual_value) AS actual_value,
nested_table.[time_period] AS time_period,
nested_table.[time_period_utc] AS time_period_utc
FROM(
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table.[target_column])
OVER (PARTITION BY
CAST(analyzed_table.[date_column] AS date),
CAST(CAST(analyzed_table.[date_column] AS date) AS DATETIME)
) AS actual_value,
CAST(analyzed_table.[date_column] AS date) AS time_period,
CAST((CAST(analyzed_table.[date_column] AS date)) AS DATETIME) AS time_period_utc
FROM [your_sql_server_database].[<target_schema>].[<target_table>] AS analyzed_table) AS nested_table
GROUP BY nested_table.[time_period], nested_table.[time_period_utc]
ORDER BY nested_table.[time_period], nested_table.[time_period_utc]
Teradata
{% import '/dialects/teradata.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}} * 1.0) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table."target_column" * 1.0) AS actual_value,
CAST(analyzed_table."date_column" AS DATE) AS time_period,
CAST(CAST(analyzed_table."date_column" AS DATE) AS TIMESTAMP) AS time_period_utc
FROM "<target_schema>"."<target_table>" analyzed_table
GROUP BY time_period, time_period_utc
ORDER BY time_period, time_period_utc
Trino
{% import '/dialects/trino.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }} {{ lib.render_time_dimension_expression(table_alias_prefix) }}
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table."time_period" AS time_period,
nested_table."time_period_utc" AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
APPROX_PERCENTILE(
CAST({{ lib.render_target_column('analyzed_table')}} AS DOUBLE),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
MAX(nested_table.actual_value) AS actual_value,
nested_table."time_period" AS time_period,
nested_table."time_period_utc" AS time_period_utc
FROM(
SELECT
APPROX_PERCENTILE(
CAST(analyzed_table."target_column" AS DOUBLE),
0.5)
OVER (PARTITION BY
CAST(analyzed_table."date_column" AS date),
CAST(analyzed_table."date_column" AS date)
) AS actual_value,
CAST(analyzed_table."date_column" AS date) AS time_period,
CAST(CAST(analyzed_table."date_column" AS date) AS TIMESTAMP) AS time_period_utc
FROM "your_trino_catalog"."<target_schema>"."<target_table>" AS analyzed_table) AS nested_table
GROUP BY time_period, time_period_utc
ORDER BY time_period, time_period_utc
Expand the Configure with data grouping section to see additional examples for configuring this data quality checks to use data grouping (GROUP BY).
Configuration with data grouping
Sample configuration with data grouping enabled (YAML) The sample below shows how to configure the data grouping and how it affects the generated SQL query.
# yaml-language-server: $schema=https://cloud.dqops.com/dqo-yaml-schema/TableYaml-schema.json
apiVersion: dqo/v1
kind: table
spec:
timestamp_columns:
partition_by_column: date_column
incremental_time_window:
daily_partitioning_recent_days: 7
monthly_partitioning_recent_months: 1
default_grouping_name: group_by_country_and_state
groupings:
group_by_country_and_state:
level_1:
source: column_value
column: country
level_2:
source: column_value
column: state
columns:
target_column:
partitioned_checks:
daily:
anomaly:
daily_partition_median_change:
parameters:
percentile_value: 0.5
warning:
max_percent: 10.0
error:
max_percent: 20.0
fatal:
max_percent: 50.0
labels:
- This is the column that is analyzed for data quality issues
date_column:
labels:
- "date or datetime column used as a daily or monthly partitioning key, dates\
\ (and times) are truncated to a day or a month by the sensor's query for\
\ partitioned checks"
country:
labels:
- column used as the first grouping key
state:
labels:
- column used as the second grouping key
Please expand the database engine name section to see the SQL query rendered by a Jinja2 template for the percentile sensor.
BigQuery
{% import '/dialects/bigquery.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }}TIMESTAMP({{ lib.render_time_dimension_expression(table_alias_prefix) }})
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table.`time_period` AS time_period,
nested_table.`time_period_utc` AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE_CONT(
({{ lib.render_target_column('analyzed_table')}}),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
MAX(nested_table.actual_value) AS actual_value,
nested_table.`time_period` AS time_period,
nested_table.`time_period_utc` AS time_period_utc,
analyzed_table.`country` AS grouping_level_1,
analyzed_table.`state` AS grouping_level_2
FROM(
SELECT
PERCENTILE_CONT(
(analyzed_table.`target_column`),
0.5)
OVER (PARTITION BY
CAST(analyzed_table.`date_column` AS DATE),
TIMESTAMP(CAST(analyzed_table.`date_column` AS DATE)),
analyzed_table.`country`,
analyzed_table.`state`
) AS actual_value,
CAST(analyzed_table.`date_column` AS DATE) AS time_period,
TIMESTAMP(CAST(analyzed_table.`date_column` AS DATE)) AS time_period_utc
FROM `your-google-project-id`.`<target_schema>`.`<target_table>` AS analyzed_table) AS nested_table
GROUP BY grouping_level_1, grouping_level_2, time_period, time_period_utc
ORDER BY grouping_level_1, grouping_level_2, time_period, time_period_utc
ClickHouse
{% import '/dialects/clickhouse.sql.jinja2' as lib with context -%}
SELECT
quantile({{ parameters.percentile_value }})({{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections_reference('analyzed_table') }}
{{- lib.render_time_dimension_projection_reference('analyzed_table') }}
FROM(
SELECT
original_table.*
{{- lib.render_data_grouping_projections('original_table') }}
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
ORDER BY {{ lib.render_target_column('original_table')}}
) analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
quantile(0.5)(analyzed_table."target_column") AS actual_value,
analyzed_table.grouping_level_1,
analyzed_table.grouping_level_2,
time_period,
time_period_utc
FROM(
SELECT
original_table.*,
original_table."country" AS grouping_level_1,
original_table."state" AS grouping_level_2,
CAST(original_table."date_column" AS DATE) AS time_period,
toDateTime64(CAST(original_table."date_column" AS DATE), 3) AS time_period_utc
FROM "<target_schema>"."<target_table>" original_table
ORDER BY original_table."target_column"
) analyzed_table
GROUP BY grouping_level_1, grouping_level_2, time_period, time_period_utc
ORDER BY grouping_level_1, grouping_level_2, time_period, time_period_utc
Databricks
{% import '/dialects/databricks.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }}TIMESTAMP({{ lib.render_time_dimension_expression(table_alias_prefix) }})
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table.`time_period` AS time_period,
nested_table.`time_period_utc` AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE(
({{ lib.render_target_column('analyzed_table')}}),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
MAX(nested_table.actual_value) AS actual_value,
nested_table.`time_period` AS time_period,
nested_table.`time_period_utc` AS time_period_utc,
analyzed_table.`country` AS grouping_level_1,
analyzed_table.`state` AS grouping_level_2
FROM(
SELECT
PERCENTILE(
(analyzed_table.`target_column`),
0.5)
OVER (PARTITION BY
CAST(analyzed_table.`date_column` AS DATE),
TIMESTAMP(CAST(analyzed_table.`date_column` AS DATE)),
analyzed_table.`country`,
analyzed_table.`state`
) AS actual_value,
CAST(analyzed_table.`date_column` AS DATE) AS time_period,
TIMESTAMP(CAST(analyzed_table.`date_column` AS DATE)) AS time_period_utc
FROM `<target_schema>`.`<target_table>` AS analyzed_table) AS nested_table
GROUP BY grouping_level_1, grouping_level_2, time_period, time_period_utc
ORDER BY grouping_level_1, grouping_level_2, time_period, time_period_utc
DB2
{% import '/dialects/db2.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections_reference('analyzed_table') }}
{{- lib.render_time_dimension_projection_reference('analyzed_table') }}
FROM(
SELECT
original_table.*
{{- lib.render_data_grouping_projections('original_table') }}
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
) analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table."target_column") AS actual_value,
analyzed_table.grouping_level_1,
analyzed_table.grouping_level_2,
time_period,
time_period_utc
FROM(
SELECT
original_table.*,
original_table."country" AS grouping_level_1,
original_table."state" AS grouping_level_2,
CAST(original_table."date_column" AS DATE) AS time_period,
TIMESTAMP(CAST(original_table."date_column" AS DATE)) AS time_period_utc
FROM "<target_schema>"."<target_table>" original_table
) analyzed_table
GROUP BY grouping_level_1, grouping_level_2, time_period, time_period_utc
ORDER BY grouping_level_1, grouping_level_2, time_period, time_period_utc
DuckDB
{% import '/dialects/duckdb.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table."target_column") AS actual_value,
analyzed_table."country" AS grouping_level_1,
analyzed_table."state" AS grouping_level_2,
CAST(analyzed_table."date_column" AS date) AS time_period,
CAST((CAST(analyzed_table."date_column" AS date)) AS TIMESTAMP WITH TIME ZONE) AS time_period_utc
FROM AS analyzed_table
GROUP BY grouping_level_1, grouping_level_2, time_period, time_period_utc
ORDER BY grouping_level_1, grouping_level_2, time_period, time_period_utc
HANA
{% import '/dialects/hana.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }} {{ lib.render_time_dimension_expression(table_alias_prefix) }}
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(analyzed_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
time_period,
time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('original_table')}})
OVER (
{%- if lib.data_groupings is not none or lib.time_series is not none %}
PARTITION BY
{%- endif -%}
{{ render_local_time_dimension_projection('original_table') -}}
{{ render_local_data_grouping_projections('original_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
{{- lib.render_where_clause(indentation = ' ', table_alias_prefix='original_table') -}}
) analyzed_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
MAX(analyzed_table.actual_value) AS actual_value,
time_period,
time_period_utc,
analyzed_table."country" AS grouping_level_1,
analyzed_table."state" AS grouping_level_2
FROM(
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY original_table."target_column")
OVER (
PARTITION BY
CAST(original_table."date_column" AS DATE),
CAST(original_table."date_column" AS DATE),
original_table."country",
original_table."state"
) AS actual_value,
CAST(original_table."date_column" AS DATE) AS time_period,
TO_TIMESTAMP(CAST(original_table."date_column" AS DATE)) AS time_period_utc
FROM "<target_schema>"."<target_table>" original_table) analyzed_table
GROUP BY grouping_level_1, grouping_level_2, time_period, time_period_utc
ORDER BY grouping_level_1, grouping_level_2, time_period, time_period_utc
Oracle
{% import '/dialects/oracle.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections_reference('analyzed_table') }}
{{- lib.render_time_dimension_projection_reference('analyzed_table') }}
FROM(
SELECT
original_table.*
{{- lib.render_data_grouping_projections('original_table') }}
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
) analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table."target_column") AS actual_value,
analyzed_table.grouping_level_1,
analyzed_table.grouping_level_2
,
time_period,
time_period_utc
FROM(
SELECT
original_table.*,
original_table."country" AS grouping_level_1,
original_table."state" AS grouping_level_2,
TRUNC(CAST(original_table."date_column" AS DATE)) AS time_period,
CAST(TRUNC(CAST(original_table."date_column" AS DATE)) AS TIMESTAMP WITH TIME ZONE) AS time_period_utc
FROM "<target_schema>"."<target_table>" original_table
) analyzed_table
GROUP BY grouping_level_1, grouping_level_2, time_period, time_period_utc
ORDER BY grouping_level_1, grouping_level_2, time_period, time_period_utc
PostgreSQL
{% import '/dialects/postgresql.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table."target_column") AS actual_value,
analyzed_table."country" AS grouping_level_1,
analyzed_table."state" AS grouping_level_2,
CAST(analyzed_table."date_column" AS date) AS time_period,
CAST((CAST(analyzed_table."date_column" AS date)) AS TIMESTAMP WITH TIME ZONE) AS time_period_utc
FROM "your_postgresql_database"."<target_schema>"."<target_table>" AS analyzed_table
GROUP BY grouping_level_1, grouping_level_2, time_period, time_period_utc
ORDER BY grouping_level_1, grouping_level_2, time_period, time_period_utc
Presto
{% import '/dialects/presto.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }} {{ lib.render_time_dimension_expression(table_alias_prefix) }}
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table."time_period" AS time_period,
nested_table."time_period_utc" AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
APPROX_PERCENTILE(
CAST({{ lib.render_target_column('analyzed_table')}} AS DOUBLE),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
MAX(nested_table.actual_value) AS actual_value,
nested_table."time_period" AS time_period,
nested_table."time_period_utc" AS time_period_utc,
analyzed_table."country" AS grouping_level_1,
analyzed_table."state" AS grouping_level_2
FROM(
SELECT
APPROX_PERCENTILE(
CAST(analyzed_table."target_column" AS DOUBLE),
0.5)
OVER (PARTITION BY
CAST(analyzed_table."date_column" AS date),
CAST(analyzed_table."date_column" AS date),
analyzed_table."country",
analyzed_table."state"
) AS actual_value,
CAST(analyzed_table."date_column" AS date) AS time_period,
CAST(CAST(analyzed_table."date_column" AS date) AS TIMESTAMP) AS time_period_utc
FROM "your_trino_database"."<target_schema>"."<target_table>" AS analyzed_table) AS nested_table
GROUP BY grouping_level_1, grouping_level_2, time_period, time_period_utc
ORDER BY grouping_level_1, grouping_level_2, time_period, time_period_utc
QuestDB
{% import '/dialects/questdb.sql.jinja2' as lib with context -%}
SELECT
APPROX_PERCENTILE({{ lib.render_target_column('analyzed_table')}} * 1.0, {{ parameters.percentile_value }}, 2) AS actual_value
{{- lib.render_data_grouping_projections_reference('analyzed_table') }}
{{- lib.render_time_dimension_projection_reference('analyzed_table') }}
FROM(
SELECT
original_table.*
{{- lib.render_data_grouping_projections('original_table') }}
{{- lib.render_time_dimension_projection('original_table') }}
FROM {{ lib.render_target_table() }} original_table
) analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
APPROX_PERCENTILE(analyzed_table."target_column" * 1.0, 0.5, 2) AS actual_value,
analyzed_table.grouping_level_1,
analyzed_table.grouping_level_2,
time_period,
time_period_utc
FROM(
SELECT
original_table.*,
original_table."country" AS grouping_level_1,
original_table."state" AS grouping_level_2,
CAST(DATE_TRUNC('day', original_table."date_column") AS DATE) AS time_period,
CAST((CAST(DATE_TRUNC('day', original_table."date_column") AS DATE)) AS TIMESTAMP WITH TIME ZONE) AS time_period_utc
FROM "<target_table>" original_table
) analyzed_table
GROUP BY grouping_level_1, grouping_level_2, time_period, time_period_utc
ORDER BY grouping_level_1, grouping_level_2, time_period, time_period_utc
Redshift
{% import '/dialects/redshift.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table."target_column") AS actual_value,
analyzed_table."country" AS grouping_level_1,
analyzed_table."state" AS grouping_level_2,
CAST(analyzed_table."date_column" AS date) AS time_period,
CAST((CAST(analyzed_table."date_column" AS date)) AS TIMESTAMP WITH TIME ZONE) AS time_period_utc
FROM "your_redshift_database"."<target_schema>"."<target_table>" AS analyzed_table
GROUP BY grouping_level_1, grouping_level_2, time_period, time_period_utc
ORDER BY grouping_level_1, grouping_level_2, time_period, time_period_utc
Snowflake
{% import '/dialects/snowflake.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}}) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table."target_column") AS actual_value,
analyzed_table."country" AS grouping_level_1,
analyzed_table."state" AS grouping_level_2,
CAST(analyzed_table."date_column" AS date) AS time_period,
TO_TIMESTAMP(CAST(analyzed_table."date_column" AS date)) AS time_period_utc
FROM "your_snowflake_database"."<target_schema>"."<target_table>" AS analyzed_table
GROUP BY grouping_level_1, grouping_level_2, time_period, time_period_utc
ORDER BY grouping_level_1, grouping_level_2, time_period, time_period_utc
Spark
{% import '/dialects/spark.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }}TIMESTAMP({{ lib.render_time_dimension_expression(table_alias_prefix) }})
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table.`time_period` AS time_period,
nested_table.`time_period_utc` AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE(
({{ lib.render_target_column('analyzed_table')}}),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
MAX(nested_table.actual_value) AS actual_value,
nested_table.`time_period` AS time_period,
nested_table.`time_period_utc` AS time_period_utc,
analyzed_table.`country` AS grouping_level_1,
analyzed_table.`state` AS grouping_level_2
FROM(
SELECT
PERCENTILE(
(analyzed_table.`target_column`),
0.5)
OVER (PARTITION BY
CAST(analyzed_table.`date_column` AS DATE),
TIMESTAMP(CAST(analyzed_table.`date_column` AS DATE)),
analyzed_table.`country`,
analyzed_table.`state`
) AS actual_value,
CAST(analyzed_table.`date_column` AS DATE) AS time_period,
TIMESTAMP(CAST(analyzed_table.`date_column` AS DATE)) AS time_period_utc
FROM `<target_schema>`.`<target_table>` AS analyzed_table) AS nested_table
GROUP BY grouping_level_1, grouping_level_2, time_period, time_period_utc
ORDER BY grouping_level_1, grouping_level_2, time_period, time_period_utc
SQL Server
{% import '/dialects/sqlserver.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }}CAST({{ lib.render_time_dimension_expression(table_alias_prefix) }} AS DATETIME)
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
{%- macro render_time_period_columns() -%}
{% if lib.time_series is not none -%}
nested_table.[time_period], nested_table.[time_period_utc]
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table.[time_period] AS time_period,
nested_table.[time_period_utc] AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{% if lib.time_series is not none or (data_groupings is not none and (data_groupings | length()) > 0) -%}
GROUP BY {{render_time_period_columns()}} {{- lib.render_data_grouping_projections('analyzed_table', set_leading_comma=(lib.time_series is not none)) }}
ORDER BY {{render_time_period_columns()}} {{- lib.render_data_grouping_projections('analyzed_table', set_leading_comma=(lib.time_series is not none)) }}
{%- endif -%}
SELECT
MAX(nested_table.actual_value) AS actual_value,
nested_table.[time_period] AS time_period,
nested_table.[time_period_utc] AS time_period_utc,
analyzed_table.[country] AS grouping_level_1,
analyzed_table.[state] AS grouping_level_2
FROM(
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table.[target_column])
OVER (PARTITION BY
CAST(analyzed_table.[date_column] AS date),
CAST(CAST(analyzed_table.[date_column] AS date) AS DATETIME),
analyzed_table.[country],
analyzed_table.[state]
) AS actual_value,
CAST(analyzed_table.[date_column] AS date) AS time_period,
CAST((CAST(analyzed_table.[date_column] AS date)) AS DATETIME) AS time_period_utc
FROM [your_sql_server_database].[<target_schema>].[<target_table>] AS analyzed_table) AS nested_table
GROUP BY nested_table.[time_period], nested_table.[time_period_utc],
analyzed_table.[country] AS grouping_level_1,
analyzed_table.[state] AS grouping_level_2
ORDER BY nested_table.[time_period], nested_table.[time_period_utc],
analyzed_table.[country] AS grouping_level_1,
analyzed_table.[state] AS grouping_level_2
Teradata
{% import '/dialects/teradata.sql.jinja2' as lib with context -%}
SELECT
PERCENTILE_CONT({{ parameters.percentile_value }})
WITHIN GROUP (ORDER BY {{ lib.render_target_column('analyzed_table')}} * 1.0) AS actual_value
{{- lib.render_data_grouping_projections('analyzed_table') }}
{{- lib.render_time_dimension_projection('analyzed_table') }}
FROM {{ lib.render_target_table() }} analyzed_table
{{- lib.render_where_clause() -}}
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
PERCENTILE_CONT(0.5)
WITHIN GROUP (ORDER BY analyzed_table."target_column" * 1.0) AS actual_value,
analyzed_table."country" AS grouping_level_1,
analyzed_table."state" AS grouping_level_2,
CAST(analyzed_table."date_column" AS DATE) AS time_period,
CAST(CAST(analyzed_table."date_column" AS DATE) AS TIMESTAMP) AS time_period_utc
FROM "<target_schema>"."<target_table>" analyzed_table
GROUP BY grouping_level_1, grouping_level_2, time_period, time_period_utc
ORDER BY grouping_level_1, grouping_level_2, time_period, time_period_utc
Trino
{% import '/dialects/trino.sql.jinja2' as lib with context -%}
{%- macro render_local_time_dimension_projection(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.time_series is not none -%}
{{- lib.eol() -}}
{{ indentation }}{{ lib.render_time_dimension_expression(table_alias_prefix) }},{{ lib.eol() -}}
{{ indentation }} {{ lib.render_time_dimension_expression(table_alias_prefix) }}
{{- "," if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -}}
{%- endif -%}
{%- endmacro -%}
{%- macro render_local_data_grouping_projections(table_alias_prefix = 'analyzed_table', indentation = ' ') -%}
{%- if lib.data_groupings is not none and (lib.data_groupings | length()) > 0 -%}
{%- for attribute in lib.data_groupings -%}
{%- with data_grouping_level = lib.data_groupings[attribute] -%}
{%- if data_grouping_level.source == 'tag' -%}
{{ lib.eol() }}{{ indentation }}{{ lib.make_text_constant(data_grouping_level.tag) }}
{%- elif data_grouping_level.source == 'column_value' -%}
{{ lib.eol() }}{{ indentation }}{{ table_alias_prefix }}.{{ lib.quote_identifier(data_grouping_level.column) }}
{%- endif -%}
{{ "," if not loop.last }}
{%- endwith %}
{%- endfor -%}
{%- endif -%}
{%- endmacro -%}
SELECT
MAX(nested_table.actual_value) AS actual_value {{-"," if lib.time_series is not none -}}
{% if lib.time_series is not none %}
nested_table."time_period" AS time_period,
nested_table."time_period_utc" AS time_period_utc
{%- endif -%}
{{- lib.render_data_grouping_projections('analyzed_table') }}
FROM(
SELECT
APPROX_PERCENTILE(
CAST({{ lib.render_target_column('analyzed_table')}} AS DOUBLE),
{{ parameters.percentile_value }})
OVER (PARTITION BY
{%- if lib.data_groupings is none and lib.time_series is none %}
NULL
{%- endif -%}
{{render_local_time_dimension_projection('analyzed_table') -}}
{{render_local_data_grouping_projections('analyzed_table') }}
) AS actual_value
{{- lib.render_time_dimension_projection('analyzed_table', indentation=' ') }}
FROM {{ lib.render_target_table() }} AS analyzed_table
{{- lib.render_where_clause(indentation = ' ') -}}) AS nested_table
{{- lib.render_group_by() -}}
{{- lib.render_order_by() -}}
SELECT
MAX(nested_table.actual_value) AS actual_value,
nested_table."time_period" AS time_period,
nested_table."time_period_utc" AS time_period_utc,
analyzed_table."country" AS grouping_level_1,
analyzed_table."state" AS grouping_level_2
FROM(
SELECT
APPROX_PERCENTILE(
CAST(analyzed_table."target_column" AS DOUBLE),
0.5)
OVER (PARTITION BY
CAST(analyzed_table."date_column" AS date),
CAST(analyzed_table."date_column" AS date),
analyzed_table."country",
analyzed_table."state"
) AS actual_value,
CAST(analyzed_table."date_column" AS date) AS time_period,
CAST(CAST(analyzed_table."date_column" AS date) AS TIMESTAMP) AS time_period_utc
FROM "your_trino_catalog"."<target_schema>"."<target_table>" AS analyzed_table) AS nested_table
GROUP BY grouping_level_1, grouping_level_2, time_period, time_period_utc
ORDER BY grouping_level_1, grouping_level_2, time_period, time_period_utc
What's next
- Learn how to configure data quality checks in DQOps
- Look at the examples of running data quality checks, targeting tables and columns