fix: chaoss metrics aggregation issues (#4652)

* feat: add entity types to timeseries metrics and enhance contributor absence factor tests

* feat(tests): add multiple artifacts test cases for burstiness and self-merge rates

* fix(sql): correct aggregation logic in self merge rates calculation

* fix(sqlmesh): update variance calculation to ensure correct aggregation in burstiness metrics

---------

Co-authored-by: ken <[email protected]>

Author: kenyiu

warehouse/oso_sqlmesh/models/metrics_factories.py

@@ -833,6 +833,7 @@ def no_gaps_audit_factory(config: MetricQueryConfig) -> tuple[str, dict] | None:
                 "biannually",
                 "yearly",
             ],
+            entity_types=["artifact", "project", "collection"],
             over_all_time=True,
             metadata=MetricMetadata(
                 display_name="Contributor Absence Factor",

warehouse/oso_sqlmesh/oso_metrics/code/burstiness.sql

@@ -52,13 +52,14 @@ variance_calc as (
   from activity_stats
 )
 select
-  variance_calc.metrics_sample_date as metrics_sample_date,
-  variance_calc.event_source as event_source,
-  variance_calc.to_artifact_id as to_artifact_id,
+  variance_calc.metrics_sample_date,
+  variance_calc.event_source,
+  variance_calc.to_artifact_id,
   '' as from_artifact_id,
   @metric_name() as metric,
   cast(case
-    when mean_activity > 0 then variance_activity * variance_activity / mean_activity
+    when sum(mean_activity) > 0 then sum(variance_activity * variance_activity) / sum(mean_activity)
     else NULL
   end as DOUBLE) as amount
 from variance_calc
+group by 1, 2, 3, 4, 5

warehouse/oso_sqlmesh/oso_metrics/code/contributor_absence_factor.sql

@@ -35,7 +35,7 @@ ranked_contributors as (
     cc.contribution_amount,
     tc.total_amount,
     row_number() over (
-      partition by cc.metrics_sample_date, cc.event_source, cc.to_artifact_id 
+      partition by cc.metrics_sample_date, cc.event_source, cc.to_artifact_id
       order by cc.contribution_amount desc
     ) as contributor_rank
   from contributor_contributions cc
@@ -55,29 +55,20 @@ cumulative_contributions as (
     total_amount,
     contributor_rank,
     sum(contribution_amount) over (
-      partition by metrics_sample_date, event_source, to_artifact_id 
+      partition by metrics_sample_date, event_source, to_artifact_id
       order by contributor_rank
       rows unbounded preceding
     ) as cumulative_amount
   from ranked_contributors
-),
-
-absence_factor as (
-  select
-    metrics_sample_date,
-    event_source,
-    to_artifact_id,
-    min(contributor_rank) as contributor_absence_factor
-  from cumulative_contributions
-  where cumulative_amount >= (total_amount * 0.5)
-  group by 1, 2, 3
 )
 
 select
-  af.metrics_sample_date,
-  af.event_source,
-  af.to_artifact_id,
+  metrics_sample_date,
+  event_source,
+  to_artifact_id,
   '' as from_artifact_id,
   @metric_name() as metric,
-  af.contributor_absence_factor as amount
-from absence_factor af
+  min(contributor_rank) as amount
+from cumulative_contributions
+where cumulative_amount >= (total_amount * 0.5)
+group by 1, 2, 3, 4, 5

warehouse/oso_sqlmesh/oso_metrics/code/self_merge_rates.sql

@@ -64,18 +64,19 @@ total_merged_prs as (
     to_artifact_id
 )
 select
-  total.metrics_sample_date as metrics_sample_date,
-  total.event_source as event_source,
-  total.to_artifact_id as to_artifact_id,
+  total.metrics_sample_date,
+  total.event_source,
+  total.to_artifact_id,
   '' as from_artifact_id,
   @metric_name() as metric,
-  cast(case 
-    when total.total_merged_count > 0 then 
-      coalesce(self_merged.self_merged_count, 0) / cast(total.total_merged_count as double)
-    else NULL
+  cast(case
+    when sum(total.total_merged_count) > 0
+      then sum(coalesce(self_merged.self_merged_count, 0)) / cast(sum(total.total_merged_count) as double)
+    else null
   end as double) as amount
 from total_merged_prs as total
 left join self_merged_prs_without_comments as self_merged
   on total.metrics_sample_date = self_merged.metrics_sample_date
   and total.event_source = self_merged.event_source
   and total.to_artifact_id = self_merged.to_artifact_id
+group by 1, 2, 3, 4, 5

warehouse/oso_sqlmesh/tests/test_burstiness_to_project_daily.yml

@@ -420,3 +420,74 @@ test_burstiness_to_project_daily_complex_multi_user_events:
         from_artifact_id: ''
         metric: burstiness_daily
         amount: 5.671296296
+
+
+test_burstiness_to_project_daily_multiple_artifacts_in_project:
+  model: oso.burstiness_to_project_daily
+  vars:
+    start: 2024-01-01
+    end: 2024-01-02
+  inputs:
+    oso.int_events_daily__github:
+      rows:
+      # Day 1
+      - to_artifact_id: repo_1
+        from_artifact_id: contributor_1
+        event_source: SOURCE_PROVIDER
+        event_type: COMMIT_CODE
+        bucket_day: 2024-01-01
+        amount: 100
+      - to_artifact_id: repo_2
+        from_artifact_id: contributor_2
+        event_source: SOURCE_PROVIDER
+        event_type: COMMIT_CODE
+        bucket_day: 2024-01-01
+        amount: 100
+      # Day 2
+      - to_artifact_id: repo_1
+        from_artifact_id: contributor_1
+        event_source: SOURCE_PROVIDER
+        event_type: COMMIT_CODE
+        bucket_day: 2024-01-02
+        amount: 100
+      - to_artifact_id: repo_2
+        from_artifact_id: contributor_2
+        event_source: SOURCE_PROVIDER
+        event_type: COMMIT_CODE
+        bucket_day: 2024-01-02
+        amount: 100
+    oso.int_first_last_commit_to_github_repository:
+      rows:
+      - artifact_id: repo_1
+        artifact_source_id: github
+        artifact_namespace: test
+        artifact_name: repo_1
+        first_commit_time: 2024-01-01 00:00:00
+        last_commit_time: 2024-01-02 00:00:00
+      - artifact_id: repo_2
+        artifact_source_id: github
+        artifact_namespace: test
+        artifact_name: repo_2
+        first_commit_time: 2024-01-01 00:00:00
+        last_commit_time: 2024-01-02 00:00:00
+    oso.artifacts_by_project_v1:
+      rows:
+      - artifact_id: repo_1
+        project_id: project_1
+      - artifact_id: repo_2
+        project_id: project_1
+  outputs:
+    query:
+      rows:
+      - metrics_sample_date: 2024-01-01
+        event_source: SOURCE_PROVIDER
+        to_project_id: project_1
+        from_artifact_id: ''
+        metric: burstiness_daily
+        amount: 0
+      - metrics_sample_date: 2024-01-02
+        event_source: SOURCE_PROVIDER
+        to_project_id: project_1
+        from_artifact_id: ''
+        metric: burstiness_daily
+        amount: 0

warehouse/oso_sqlmesh/tests/test_contributor_absence_factor_to_project_daily.yml

@@ -209,3 +209,40 @@ test_contributor_absence_factor_to_project_daily_no_activity:
   outputs:
     query:
       rows: []
+
+
+test_contributor_absence_factor_to_project_daily_multiple_artifacts_in_project:
+  model: oso.contributor_absence_factor_to_project_daily
+  vars:
+    start: 2024-01-01
+    end: 2024-01-01
+  inputs:
+    oso.int_events_daily__github:
+      rows:
+      - to_artifact_id: repo_1
+        from_artifact_id: contributor_1
+        event_source: SOURCE_PROVIDER
+        event_type: COMMIT_CODE
+        bucket_day: 2024-01-01
+        amount: 100
+      - to_artifact_id: repo_2
+        from_artifact_id: contributor_2
+        event_source: SOURCE_PROVIDER
+        event_type: COMMIT_CODE
+        bucket_day: 2024-01-01
+        amount: 100
+    oso.artifacts_by_project_v1:
+      rows:
+      - artifact_id: repo_1
+        project_id: project_1
+      - artifact_id: repo_2
+        project_id: project_1
+  outputs:
+    query:
+      rows:
+      - metrics_sample_date: 2024-01-01
+        event_source: SOURCE_PROVIDER
+        to_project_id: project_1
+        from_artifact_id: ''
+        metric: contributor_absence_factor_daily
+        amount: 1

warehouse/oso_sqlmesh/tests/test_self_merge_rates_to_project_daily.yml

@@ -70,7 +70,7 @@ test_self_merge_rates_to_project_daily:
         project_id: project_1
   outputs:
     query:
-      rows: 
+      rows:
       - metrics_sample_date: 2024-01-01
         event_source: GITHUB
         to_project_id: project_1
@@ -89,3 +89,110 @@ test_self_merge_rates_to_project_daily:
         from_artifact_id: ''
         metric: self_merge_rates_daily
         amount: 0
+
+
+test_self_merge_rates_to_project_daily_multiple_artifacts_in_project:
+  model: oso.self_merge_rates_to_project_daily
+  vars:
+    start: 2024-01-01
+    end: 2024-01-02
+  inputs:
+    oso.int_events_aux_prs:
+      rows:
+      # repo_1, Day 1, self-merged
+      - time: "2024-01-01 09:00:00"
+        to_artifact_id: repo_1
+        from_artifact_id: user_1
+        pr_id: pr_1_1
+        pr_number: 1
+        event_type: PULL_REQUEST_OPENED
+        event_source: GITHUB
+        merged_at: null
+        comments: 0
+      - time: "2024-01-01 10:00:00"
+        to_artifact_id: repo_1
+        from_artifact_id: user_1
+        pr_id: pr_1_1
+        pr_number: 1
+        event_type: PULL_REQUEST_MERGED
+        event_source: GITHUB
+        merged_at: "2024-01-01 10:00:00"
+        comments: 0
+      # repo_2, Day 1, self-merged
+      - time: "2024-01-01 11:00:00"
+        to_artifact_id: repo_2
+        from_artifact_id: user_2
+        pr_id: pr_2_1
+        pr_number: 1
+        event_type: PULL_REQUEST_OPENED
+        event_source: GITHUB
+        merged_at: null
+        comments: 0
+      - time: "2024-01-01 12:00:00"
+        to_artifact_id: repo_2
+        from_artifact_id: user_2
+        pr_id: pr_2_1
+        pr_number: 1
+        event_type: PULL_REQUEST_MERGED
+        event_source: GITHUB
+        merged_at: "2024-01-01 12:00:00"
+        comments: 0
+      # repo_1, Day 2, not self-merged
+      - time: "2024-01-02 09:00:00"
+        to_artifact_id: repo_1
+        from_artifact_id: user_1
+        pr_id: pr_1_2
+        pr_number: 2
+        event_type: PULL_REQUEST_OPENED
+        event_source: GITHUB
+        merged_at: null
+        comments: 0
+      - time: "2024-01-02 10:00:00"
+        to_artifact_id: repo_1
+        from_artifact_id: user_3 # Merged by another user
+        pr_id: pr_1_2
+        pr_number: 2
+        event_type: PULL_REQUEST_MERGED
+        event_source: GITHUB
+        merged_at: "2024-01-02 10:00:00"
+        comments: 0
+      # repo_2, Day 2, self-merged
+      - time: "2024-01-02 11:00:00"
+        to_artifact_id: repo_2
+        from_artifact_id: user_2
+        pr_id: pr_2_2
+        pr_number: 2
+        event_type: PULL_REQUEST_OPENED
+        event_source: GITHUB
+        merged_at: null
+        comments: 0
+      - time: "2024-01-02 12:00:00"
+        to_artifact_id: repo_2
+        from_artifact_id: user_2
+        pr_id: pr_2_2
+        pr_number: 2
+        event_type: PULL_REQUEST_MERGED
+        event_source: GITHUB
+        merged_at: "2024-01-02 12:00:00"
+        comments: 0
+    oso.artifacts_by_project_v1:
+      rows:
+      - artifact_id: repo_1
+        project_id: project_1
+      - artifact_id: repo_2
+        project_id: project_1
+  outputs:
+    query:
+      rows:
+      - metrics_sample_date: 2024-01-01
+        event_source: GITHUB
+        to_project_id: project_1
+        from_artifact_id: ''
+        metric: self_merge_rates_daily
+        amount: 1.0
+      - metrics_sample_date: 2024-01-02
+        event_source: GITHUB
+        to_project_id: project_1
+        from_artifact_id: ''
+        metric: self_merge_rates_daily
+        amount: 0.5