Cherry Pick: not null proportion schema test (#411)

* Add not_null_proportion schema test and related integration tests

* Update CHANGELOG

* Fix csv formatting and numeric typecasting

Co-authored-by: Simo Tumelius <simo.tumelius@gmail.com>

Author: jasnonaz

CHANGELOG.md

@@ -56,6 +56,7 @@ If you were relying on the position to match up your optional arguments, this ma
 ## Features
 * Add new argument, `order_by`, to `get_column_values` (code originally in [#289](https://github.com/fishtown-analytics/dbt-utils/pull/289/) from [@clausherther](https://github.com/clausherther), merged via [#349](https://github.com/fishtown-analytics/dbt-utils/pull/349/))
 * Add `slugify` macro, and use it in the pivot macro. :rotating_light: This macro uses the `re` module, which is only available in dbt v0.19.0+. As a result, this feature introduces a breaking change. ([#314](https://github.com/fishtown-analytics/dbt-utils/pull/314))
+* Add `not_null_proportion` schema test that allows the user to specify the minimum (`at_least`) tolerated proportion (e.g., `0.95`) of non-null values
 
 ## Under the hood
 * Update the default implementation of concat macro to use `||` operator ([#373](https://github.com/fishtown-analytics/dbt-utils/pull/314) from [@ChristopheDuong](https://github.com/ChristopheDuong)). Note this may be a breaking change for adapters that support `concat()` but not `||`, such as Apache Spark.

README.md

@@ -16,6 +16,7 @@ Check [dbt Hub](https://hub.getdbt.com/fishtown-analytics/dbt_utils/latest/) for
   - [cardinality_equality](#cardinality_equality-source)
   - [unique_where](#unique_where-source)
   - [not_null_where](#not_null_where-source)
+  - [not_null_proportion](#not_null_proportion-source)
   - [relationships_where](#relationships_where-source)
   - [mutually_exclusive_ranges](#mutually_exclusive_ranges-source)
   - [unique_combination_of_columns](#unique_combination_of_columns-source)
@@ -252,6 +253,22 @@ models:
               where: "_deleted = false"
 ```
 
+#### not_null_proportion ([source](macros/schema_tests/not_null_proportion.sql))
+This test validates that the proportion of non-null values present in a column is between a specified range [`at_least`, `at_most`] where `at_most` is an optional argument (default: `1.0`).
+
+**Usage:**
+```yaml
+version: 2
+
+models:
+  - name: my_model
+    columns:
+      - name: id
+        tests:
+          - dbt_utils.not_null_proportion:
+              at_least: 0.95
+```
+
 #### not_accepted_values ([source](macros/schema_tests/not_accepted_values.sql))
 This test validates that there are no rows that match the given values.
 

integration_tests/data/schema_tests/data_not_null_proportion.csv

@@ -0,0 +1,11 @@
+point_5,point_9
+1,1
+,2
+,3
+4,4
+5,5
+6,6
+,7
+,8
+,
+10,10

integration_tests/models/schema_tests/schema.yml

@@ -157,3 +157,14 @@ models:
               inclusive: true
               where: "id <> -1"
 
+  - name: data_not_null_proportion
+    columns:
+      - name: point_5
+        tests:
+          - dbt_utils.not_null_proportion:
+              at_least: 0.5
+              at_most: 0.5
+      - name: point_9
+        tests:
+          - dbt_utils.not_null_proportion:
+              at_least: 0.9

macros/schema_tests/not_null_proportion.sql

@@ -0,0 +1,26 @@
+{% macro test_not_null_proportion(model) %}
+  {{ return(adapter.dispatch('test_not_null_proportion', packages = dbt_utils._get_utils_namespaces())(model, **kwargs)) }}
+{% endmacro %}
+
+{% macro default__test_not_null_proportion(model) %}
+
+{% set column_name = kwargs.get('column_name', kwargs.get('arg')) %}
+{% set at_least = kwargs.get('at_least', kwargs.get('arg')) %}
+{% set at_most = kwargs.get('at_most', kwargs.get('arg', 1)) %}
+
+with validation as (
+  select 
+    sum(case when {{ column_name }} is null then 0 else 1 end) / cast(count(*) as numeric) as not_null_proportion
+  from {{ model }}
+),
+validation_errors as (
+  select
+    not_null_proportion
+  from validation
+  where not_null_proportion < {{ at_least }} or not_null_proportion > {{ at_most }}
+)
+select 
+  count(*)
+from validation_errors
+
+{% endmacro %}