tests for conversion metrics

dbt-labs · Jan 26, 2023 · 0d20f37 · 0d20f37
1 parent d2aba9a
commit 0d20f37
Show file tree

Hide file tree

Showing 9 changed files with 437 additions and 2 deletions.
diff --git a/metricflow/test/fixtures/model_fixtures.py b/metricflow/test/fixtures/model_fixtures.py
@@ -170,6 +170,8 @@ def template_mapping(mf_test_session_state: MetricFlowTestSessionState) -> Dict[
         "source_schema": schema,
         "accounts_source_table": f"{schema}.fct_accounts",
         "primary_accounts_table": f"{schema}.dim_primary_accounts",
+        "buys_source_table": f"{schema}.fct_buys",
+        "visits_source_table": f"{schema}.fct_visits",
     }
 
 

diff --git a/metricflow/test/fixtures/model_yamls/simple_model/data_sources/buys_source.yaml b/metricflow/test/fixtures/model_yamls/simple_model/data_sources/buys_source.yaml
@@ -0,0 +1,30 @@
+---
+data_source:
+  name: buys_source
+  description: buys_source
+  owners:
+    - [email protected]
+
+  sql_query: |
+    -- User Defined SQL Query
+    SELECT * FROM $buys_source_table
+
+  measures:
+    - name: buys
+      expr: 1
+      agg: count
+    - name: buyers
+      expr: user_id
+      agg: count_distinct
+
+  dimensions:
+    - name: ds
+      type: time
+      type_params:
+        is_primary: True
+        time_granularity: day
+
+  identifiers:
+    - name: user
+      type: foreign
+      expr: user_id
diff --git a/metricflow/test/fixtures/model_yamls/simple_model/data_sources/visits_source.yaml b/metricflow/test/fixtures/model_yamls/simple_model/data_sources/visits_source.yaml
@@ -0,0 +1,33 @@
+---
+data_source:
+  name: visits_source
+  description: visits_source
+  owners:
+    - [email protected]
+
+  sql_query: |
+    -- User Defined SQL Query
+    SELECT * FROM $visits_source_table
+
+  measures:
+    - name: visits
+      expr: 1
+      agg: count
+    - name: visitors
+      expr: user_id
+      agg: count_distinct
+
+
+  dimensions:
+    - name: ds
+      type: time
+      type_params:
+        is_primary: True
+        time_granularity: day
+    - name: referrer_id
+      type: categorical
+
+  identifiers:
+    - name: user
+      type: foreign
+      expr: user_id
diff --git a/metricflow/test/fixtures/model_yamls/simple_model/metrics.yaml b/metricflow/test/fixtures/model_yamls/simple_model/metrics.yaml
@@ -682,3 +682,31 @@ metric:
       - name: bookings
         offset_window: 5 days
         alias: bookings_5_days_ago
+---
+metric:
+  name: visit_buy_conversion_rate
+  description: conversion rate on visits-buys on a 7 day window
+  owners:
+    - [email protected]
+  type: conversion
+  type_params:
+    conversion_type_params:
+      base_measure: visits
+      conversion_measure: buys
+      window: 7 days
+      entity: user
+      calculation: conversion_rate
+---
+metric:
+  name: visit_buy_conversions
+  description: conversion count on visits-buys on a 7 day window
+  owners:
+    - [email protected]
+  type: conversion
+  type_params:
+    conversion_type_params:
+      base_measure: visits
+      conversion_measure: buys
+      window: 7 days
+      entity: user
+      calculation: conversions
diff --git a/metricflow/test/fixtures/table_fixtures.py b/metricflow/test/fixtures/table_fixtures.py
@@ -360,6 +360,61 @@ def create_simple_model_tables(mf_test_session_state: MetricFlowTestSessionState
         ),
     )
 
+    # Events data
+    visits_data = [
+        ("u0004114", "2020-01-01", "fb_ad_1"),
+        ("u0004214", "2020-01-01", "fb_ad_2"),
+        ("u0003141", "2020-01-01", "homepage_1"),
+        ("u0003154", "2020-01-01", "homepage_2"),
+        ("u1612112", "2020-01-02", "homepage_1"),
+        ("u0042324", "2020-01-02", "homepage_1"),
+        ("u0005432", "2020-01-02", "fb_ad_3"),
+        ("u0003452", "2020-01-02", "fb_ad_1"),
+        ("u0003452", "2020-01-02", "user_2"),
+        ("u0042324", "2020-01-03", "fb_ad_1"),
+        ("u0005432", "2020-01-03", "google_ad_1"),
+        ("u0005472", "2020-01-03", "google_ad_2"),
+        ("u0005414", "2020-01-04", "google_ad_1"),
+        ("u0004114", "2020-01-06", "homepage_1"),
+        ("u0004114", "2020-01-07", "fb_ad_2"),
+        ("u0004117", "2020-01-10", "google_ad_1"),
+        ("u0003141", "2020-01-12", "user_1"),
+    ]
+
+    create_table(
+        sql_client=sql_client,
+        sql_table=SqlTable(schema_name=schema, table_name="fct_visits"),
+        df=make_df(
+            sql_client=sql_client,
+            columns=["user_id", DEFAULT_DS, "referrer_id"],
+            time_columns={DEFAULT_DS},
+            data=visits_data,
+        ),
+    )
+
+    buy_data = [
+        ("u0004114", "2020-01-02"),
+        ("u0042324", "2020-01-03"),
+        ("u0042324", "2020-01-03"),
+        ("u0005432", "2020-01-04"),
+        ("u1612112", "2020-01-07"),
+        ("u0004114", "2020-01-07"),
+        ("u0004117", "2020-01-10"),
+        ("u0003141", "2020-01-07"),
+        ("u0003452", "2020-01-04"),
+    ]
+
+    create_table(
+        sql_client=sql_client,
+        sql_table=SqlTable(schema_name=schema, table_name="fct_buys"),
+        df=make_df(
+            sql_client=sql_client,
+            columns=["user_id", DEFAULT_DS],
+            time_columns={DEFAULT_DS},
+            data=buy_data,
+        ),
+    )
+
     return True
 
 

diff --git a/metricflow/test/integration/test_cases/itest_conversion_metrics.yaml b/metricflow/test/integration/test_cases/itest_conversion_metrics.yaml
@@ -0,0 +1,230 @@
+---
+integration_test:
+  name: conversion_rate_metric
+  description: Query a conversion metric that calculates the conversion rate
+  model: SIMPLE_MODEL
+  metrics: ["visit_buy_conversion_rate"]
+  group_bys: ["metric_time"]
+  check_query: |
+    SELECT
+      opportunities.metric_time AS metric_time
+      , CAST(conversions.buys AS {{ double_data_type_name }}) / CAST(NULLIF(opportunities.visits, 0) AS {{ double_data_type_name }}) AS visit_buy_conversion_rate
+    FROM (
+      SELECT
+        metric_time, SUM(a.visits) AS visits
+      FROM (
+        SELECT
+          ds AS metric_time, 1 AS visits
+        FROM {{ source_schema }}.fct_visits visits
+      ) a
+      GROUP BY
+        a.metric_time
+    ) opportunities
+    INNER JOIN (
+      SELECT
+        b.ds AS metric_time, SUM(b.buys) AS buys
+      FROM (
+        SELECT DISTINCT
+        first_value(v.ds) OVER (PARTITION BY buy_source.ds, buy_source.user_id ORDER BY v.ds DESC NULLS FIRST) AS ds
+        , first_value(v.user_id) OVER (PARTITION BY buy_source.ds, buy_source.user_id ORDER BY v.ds DESC NULLS FIRST) AS user_id
+        , first_value(v.referrer_id) OVER (PARTITION BY buy_source.ds, buy_source.user_id ORDER BY v.ds DESC NULLS FIRST) AS referrer_id
+        , buy_source.uuid
+        , 1 AS buys
+        FROM {{ source_schema }}.fct_visits v
+        INNER JOIN
+        (
+          SELECT *, {{ generate_random_uuid() }} AS uuid FROM {{ source_schema }}.fct_buys
+        ) buy_source
+        ON
+          v.user_id = buy_source.user_id AND v.ds <= buy_source.ds AND v.ds > {{ render_date_sub("buy_source", "ds", 7, TimeGranularity.DAY) }}
+      ) b
+      GROUP BY
+        b.ds
+    ) conversions
+    ON opportunities.metric_time = conversions.metric_time
+---
+integration_test:
+  name: conversion_rate_metric_with_dimension
+  description: Query a conversion metric that calculates the conversion rate without time dimension
+  model: SIMPLE_MODEL
+  metrics: ["visit_buy_conversion_rate"]
+  group_bys: ["referrer_id"]
+  check_query: |
+    SELECT
+      opportunities.referrer_id AS referrer_id
+      , CAST(conversions.buys AS {{ double_data_type_name }}) / CAST(NULLIF(opportunities.visits, 0) AS {{ double_data_type_name }}) AS visit_buy_conversion_rate
+    FROM (
+      SELECT
+        referrer_id, SUM(a.visits) AS visits
+      FROM (
+        SELECT
+          referrer_id, 1 AS visits
+        FROM {{ source_schema }}.fct_visits visits
+      ) a
+      GROUP BY
+        a.referrer_id
+    ) opportunities
+    INNER JOIN (
+      SELECT
+        referrer_id AS referrer_id, SUM(b.buys) AS buys
+      FROM (
+        SELECT DISTINCT
+        first_value(v.ds) OVER (PARTITION BY buy_source.ds, buy_source.user_id ORDER BY v.ds DESC NULLS FIRST) AS ds
+        , first_value(v.user_id) OVER (PARTITION BY buy_source.ds, buy_source.user_id ORDER BY v.ds DESC NULLS FIRST) AS user_id
+        , first_value(v.referrer_id) OVER (PARTITION BY buy_source.ds, buy_source.user_id ORDER BY v.ds DESC NULLS FIRST) AS referrer_id
+        , buy_source.uuid
+        , 1 AS buys
+        FROM {{ source_schema }}.fct_visits v
+        INNER JOIN
+        (
+          SELECT *, {{ generate_random_uuid() }} AS uuid FROM {{ source_schema }}.fct_buys
+        ) buy_source
+        ON
+          v.user_id = buy_source.user_id AND v.ds <= buy_source.ds AND v.ds > buy_source.ds - INTERVAL 7 day
+      ) b
+      GROUP BY
+        b.referrer_id
+    ) conversions
+    ON opportunities.referrer_id = conversions.referrer_id
+---
+integration_test:
+  name: conversion_rate_metric_with_multiple_dimension
+  description: Query a conversion metric that calculates the conversion rate with multiple dimension
+  model: SIMPLE_MODEL
+  metrics: ["visit_buy_conversion_rate"]
+  group_bys: ["metric_time", "referrer_id"]
+  check_query: |
+    SELECT
+      opportunities.referrer_id AS referrer_id
+      , opportunities.metric_time AS metric_time
+      , CAST(conversions.buys AS {{ double_data_type_name }}) / CAST(NULLIF(opportunities.visits, 0) AS {{ double_data_type_name }}) AS visit_buy_conversion_rate
+    FROM (
+      SELECT
+        metric_time, referrer_id, SUM(a.visits) AS visits
+      FROM (
+        SELECT
+          ds AS metric_time, referrer_id, 1 AS visits
+        FROM {{ source_schema }}.fct_visits visits
+      ) a
+      GROUP BY
+        a.referrer_id, a.metric_time
+    ) opportunities
+    INNER JOIN (
+      SELECT
+        b.ds AS metric_time, referrer_id AS referrer_id, SUM(b.buys) AS buys
+      FROM (
+        SELECT DISTINCT
+        first_value(v.ds) OVER (PARTITION BY buy_source.ds, buy_source.user_id ORDER BY v.ds DESC NULLS FIRST) AS ds
+        , first_value(v.user_id) OVER (PARTITION BY buy_source.ds, buy_source.user_id ORDER BY v.ds DESC NULLS FIRST) AS user_id
+        , first_value(v.referrer_id) OVER (PARTITION BY buy_source.ds, buy_source.user_id ORDER BY v.ds DESC NULLS FIRST) AS referrer_id
+        , buy_source.uuid
+        , 1 AS buys
+        FROM {{ source_schema }}.fct_visits v
+        INNER JOIN
+        (
+          SELECT *, {{ generate_random_uuid() }} AS uuid FROM {{ source_schema }}.fct_buys
+        ) buy_source
+        ON
+          v.user_id = buy_source.user_id AND v.ds <= buy_source.ds AND v.ds > buy_source.ds - INTERVAL 7 day
+      ) b
+      GROUP BY
+        b.referrer_id, b.ds
+    ) conversions
+    ON opportunities.referrer_id = conversions.referrer_id AND opportunities.metric_time = conversions.metric_time
+---
+integration_test:
+  name: conversion_count_metric
+  description: Query a conversion metric that calculates the conversion count
+  model: SIMPLE_MODEL
+  metrics: ["visit_buy_conversions"]
+  group_bys: ["metric_time"]
+  check_query: |
+    SELECT
+      conversions.metric_time AS metric_time
+      , conversions.buys AS visit_buy_conversions
+    FROM (
+      SELECT
+        b.ds AS metric_time, SUM(b.buys) AS buys
+      FROM (
+        SELECT DISTINCT
+        first_value(v.ds) OVER (PARTITION BY buy_source.ds, buy_source.user_id ORDER BY v.ds DESC NULLS FIRST) AS ds
+        , first_value(v.user_id) OVER (PARTITION BY buy_source.ds, buy_source.user_id ORDER BY v.ds DESC NULLS FIRST) AS user_id
+        , first_value(v.referrer_id) OVER (PARTITION BY buy_source.ds, buy_source.user_id ORDER BY v.ds DESC NULLS FIRST) AS referrer_id
+        , buy_source.uuid
+        , 1 AS buys
+        FROM {{ source_schema }}.fct_visits v
+        INNER JOIN
+        (
+          SELECT *, {{ generate_random_uuid() }} AS uuid FROM {{ source_schema }}.fct_buys
+        ) buy_source
+        ON
+          v.user_id = buy_source.user_id AND v.ds <= buy_source.ds AND v.ds > buy_source.ds - INTERVAL 7 day
+      ) b
+      GROUP BY
+        b.ds
+    ) conversions
+---
+integration_test:
+  name: conversion_count_metric_with_dimension
+  description: Query a conversion metric that calculates the conversion count without time dimension
+  model: SIMPLE_MODEL
+  metrics: ["visit_buy_conversions"]
+  group_bys: ["referrer_id"]
+  check_query: |
+    SELECT
+      conversions.referrer_id AS referrer_id
+      , conversions.buys AS visit_buy_conversions
+    FROM (
+      SELECT
+        referrer_id AS referrer_id, SUM(b.buys) AS buys
+      FROM (
+        SELECT DISTINCT
+        first_value(v.ds) OVER (PARTITION BY buy_source.ds, buy_source.user_id ORDER BY v.ds DESC NULLS FIRST) AS ds
+        , first_value(v.user_id) OVER (PARTITION BY buy_source.ds, buy_source.user_id ORDER BY v.ds DESC NULLS FIRST) AS user_id
+        , first_value(v.referrer_id) OVER (PARTITION BY buy_source.ds, buy_source.user_id ORDER BY v.ds DESC NULLS FIRST) AS referrer_id
+        , buy_source.uuid
+        , 1 AS buys
+        FROM {{ source_schema }}.fct_visits v
+        INNER JOIN
+        (
+          SELECT *, {{ generate_random_uuid() }} AS uuid FROM {{ source_schema }}.fct_buys
+        ) buy_source
+        ON
+          v.user_id = buy_source.user_id AND v.ds <= buy_source.ds AND v.ds > buy_source.ds - INTERVAL 7 day
+      ) b
+      GROUP BY
+        b.referrer_id
+    ) conversions
+---
+integration_test:
+  name: conversion_count_metric_with_multiple_dimension
+  description: Query a conversion metric that calculates the conversion count with multiple dimension
+  model: SIMPLE_MODEL
+  metrics: ["visit_buy_conversions"]
+  group_bys: ["metric_time", "referrer_id"]
+  check_query: |
+    SELECT
+      conversions.referrer_id AS referrer_id
+      , conversions.metric_time AS metric_time
+      , conversions.buys AS visit_buy_conversions
+    FROM (
+      SELECT
+        b.ds AS metric_time, referrer_id AS referrer_id, SUM(b.buys) AS buys
+      FROM (
+        SELECT DISTINCT
+        first_value(v.ds) OVER (PARTITION BY buy_source.ds, buy_source.user_id ORDER BY v.ds DESC NULLS FIRST) AS ds
+        , first_value(v.user_id) OVER (PARTITION BY buy_source.ds, buy_source.user_id ORDER BY v.ds DESC NULLS FIRST) AS user_id
+        , first_value(v.referrer_id) OVER (PARTITION BY buy_source.ds, buy_source.user_id ORDER BY v.ds DESC NULLS FIRST) AS referrer_id
+        , buy_source.uuid
+        , 1 AS buys
+        FROM {{ source_schema }}.fct_visits v
+        INNER JOIN
+        (
+          SELECT *, {{ generate_random_uuid() }} AS uuid FROM {{ source_schema }}.fct_buys
+        ) buy_source
+        ON
+          v.user_id = buy_source.user_id AND v.ds <= buy_source.ds AND v.ds > buy_source.ds - INTERVAL 7 day
+      ) b
+      GROUP BY
+        b.referrer_id, b.ds
+    ) conversions