refactor(analytics): update metrics with stat_event cohort

analytics/dbt/analytics/models/intermediate/mission/__models.yml

@@ -194,10 +194,11 @@ models:
         description: Timestamp de dernière mise à jour propagé depuis la mission/adresse.
         tests:
           - not_null
-  - name: int_mission_first_events
+  - name: int_mission_click_apply_metrics
     description: >
       Vue intermédiaire au grain mission + diffuseur + source, qui expose les premières
-      dates de click/apply et les délais depuis la création de la mission.
+      dates de click/apply, les métriques multi-candidatures par click et les délais
+      depuis la création de la mission.
     columns:
       - name: mission_id
         description: Identifiant de la mission.
@@ -207,18 +208,33 @@ models:
               arguments:
                 to: ref('mission')
                 field: id
+      - name: publisher_id
+        description: Identifiant du partenaire annonceur de la mission.
+        tests:
+          - not_null
+          - relationships:
+              arguments:
+                to: ref('publisher')
+                field: id
       - name: mission_created_at
         description: Timestamp de création de la mission (base des délais).
         tests:
           - not_null
       - name: from_publisher_id
         description: Identifiant du diffuseur à l'origine de l'événement (nullable).
+        tests:
+          - relationships:
+              arguments:
+                to: ref('publisher')
+                field: id
       - name: source
         description: Source normalisée de l'événement (api, widget, campaign).
       - name: source_id
         description: Identifiant de la source (widget/campaign/publisher selon source).
       - name: first_click_at
         description: Timestamp du premier click observé sur la mission.
+        tests:
+          - not_null
       - name: first_apply_at
         description: Timestamp de la première candidature observée sur la mission.
       - name: click_count
@@ -229,8 +245,18 @@ models:
         description: Nombre total de candidatures (apply) observées pour la mission.
         tests:
           - not_null
+      - name: click_with_apply_count
+        description: Nombre de clicks reliés à au moins une candidature.
+        tests:
+          - not_null
+      - name: click_with_multi_apply_count
+        description: Nombre de clicks reliés à au moins deux candidatures.
+        tests:
+          - not_null
       - name: conversion_rate
         description: Ratio candidatures/redirections (`apply_count / click_count`).
+      - name: multi_apply_share
+        description: Part des clicks avec plusieurs candidatures (`>=2`) parmi les clicks avec candidature.
       - name: time_to_click_secs
         description: Délai en secondes entre création de mission et premier click.
       - name: time_to_apply_secs

analytics/dbt/analytics/models/intermediate/mission/int_mission_first_events.sql → analytics/dbt/analytics/models/intermediate/mission/int_mission_click_apply_metrics.sql

@@ -4,8 +4,8 @@
   incremental_strategy = 'delete+insert',
   on_schema_change = 'sync_all_columns',
   post_hook = [
-    'create index if not exists "int_mission_first_events_mission_id_idx" on {{ this }} (mission_id)',
-    'create index if not exists "int_mission_first_events_from_publisher_id_idx" on {{ this }} (from_publisher_id)'
+    'create index if not exists "int_mission_click_apply_metrics_mission_id_idx" on {{ this }} (mission_id)',
+    'create index if not exists "int_mission_click_apply_metrics_from_publisher_id_idx" on {{ this }} (from_publisher_id)'
   ]
 ) }}
 
@@ -27,11 +27,11 @@ affected_missions as (
 
   union
 
-  select distinct ge.mission_id
-  from {{ ref('int_stat_event_global') }} as ge
+  select distinct secm.mission_id
+  from {{ ref('int_stat_event_click_metrics') }} as secm
   where
-    ge.mission_id is not null
-    and coalesce(ge.updated_at, ge.created_at)
+    secm.mission_id is not null
+    and coalesce(secm.updated_at, secm.click_created_at)
     >= (select lr.last_updated_at from last_run as lr)
 ),
 
@@ -60,21 +60,24 @@ missions as (
 
 events as (
   select
-    mission_id,
-    from_publisher_id,
-    source,
-    source_id,
-    min(created_at) filter (where type = 'click') as first_click_at,
-    min(created_at) filter (where type = 'apply') as first_apply_at,
-    count(*) filter (where type = 'click') as click_count,
-    count(*) filter (where type = 'apply') as apply_count,
-    max(updated_at) as events_updated_at
-  from {{ ref('int_stat_event_global') }}
+    secm.mission_id,
+    secm.from_publisher_id,
+    secm.source,
+    secm.source_id,
+    min(secm.click_created_at) as first_click_at,
+    min(secm.first_apply_at) as first_apply_at,
+    count(*) as click_count,
+    sum(secm.apply_count) as apply_count,
+    count(*) filter (where secm.has_apply) as click_with_apply_count,
+    count(*) filter (
+      where secm.has_multi_apply
+    ) as click_with_multi_apply_count,
+    max(secm.updated_at) as events_updated_at
+  from {{ ref('int_stat_event_click_metrics') }} as secm
   where
-    mission_id is not null
-    and type in ('click', 'apply')
+    secm.mission_id is not null
     {% if is_incremental() %}
-      and mission_id in (
+      and secm.mission_id in (
         select am.mission_id from affected_missions as am
       )
     {% endif %}
@@ -102,6 +105,8 @@ base as (
     e.first_apply_at,
     e.click_count,
     e.apply_count,
+    e.click_with_apply_count,
+    e.click_with_multi_apply_count,
     greatest(
       coalesce(m.mission_updated_at, '1900-01-01'::timestamp),
       coalesce(e.events_updated_at, '1900-01-01'::timestamp)
@@ -131,6 +136,8 @@ select
   first_apply_at,
   click_count,
   apply_count,
+  click_with_apply_count,
+  click_with_multi_apply_count,
   updated_at,
   case
     when mission_created_at is null or first_click_at is null then null
@@ -149,5 +156,9 @@ select
   case
     when click_count = 0 then null
     else apply_count::numeric / click_count
-  end as conversion_rate
+  end as conversion_rate,
+  case
+    when click_with_apply_count = 0 then null
+    else click_with_multi_apply_count::numeric / click_with_apply_count
+  end as multi_apply_share
 from base

analytics/dbt/analytics/models/intermediate/stat_event/__models.yml

@@ -0,0 +1,104 @@
+version: 2
+
+models:
+  - name: int_stat_event_global
+    description: >
+      Vue intermédiaire des événements statistiques hors bots, avec normalisation
+      des sources et des identifiants nécessaires aux agrégations downstream.
+    columns:
+      - name: stat_event_id
+        description: Identifiant unique de l'événement source.
+        tests:
+          - not_null
+          - unique
+      - name: created_at
+        description: Horodatage de création de l'événement.
+        tests:
+          - not_null
+      - name: updated_at
+        description: Horodatage de dernière mise à jour.
+        tests:
+          - not_null
+      - name: type
+        description: Type d'événement (`click`, `apply`, `print`, `account`).
+        tests:
+          - not_null
+      - name: source
+        description: Source normalisée de l'événement.
+        tests:
+          - not_null
+      - name: from_publisher_id
+        description: Identifiant du diffuseur à l'origine de l'événement.
+        tests:
+          - relationships:
+              arguments:
+                to: ref('publisher')
+                field: id
+      - name: to_publisher_id
+        description: Identifiant de l'annonceur cible de l'événement.
+        tests:
+          - relationships:
+              arguments:
+                to: ref('publisher')
+                field: id
+      - name: mission_id
+        description: Identifiant de mission associé à l'événement (nullable).
+        tests:
+          - relationships:
+              arguments:
+                to: ref('mission')
+                field: id
+
+  - name: int_stat_event_click_metrics
+    description: >
+      Vue intermédiaire au grain click (une ligne par `stat_event` de type click),
+      qui centralise les candidatures rattachées au click via `click_id`.
+    columns:
+      - name: click_stat_event_id
+        description: Identifiant du click (`stat_event.id`).
+        tests:
+          - not_null
+          - unique
+      - name: mission_id
+        description: Identifiant de la mission portée par le click.
+        tests:
+          - not_null
+          - relationships:
+              arguments:
+                to: ref('mission')
+                field: id
+      - name: from_publisher_id
+        description: Identifiant du diffuseur à l'origine du click.
+        tests:
+          - relationships:
+              arguments:
+                to: ref('publisher')
+                field: id
+      - name: source
+        description: Source normalisée des événements du grain.
+        tests:
+          - not_null
+      - name: source_id
+        description: Identifiant de la source (widget/campaign/publisher selon source).
+      - name: click_created_at
+        description: Horodatage du click.
+        tests:
+          - not_null
+      - name: first_apply_at
+        description: Horodatage de la première candidature liée au click.
+      - name: apply_count
+        description: Nombre total de candidatures liées au click.
+        tests:
+          - not_null
+      - name: has_apply
+        description: Indique si le click est relié à au moins une candidature.
+        tests:
+          - not_null
+      - name: has_multi_apply
+        description: Indique si le click est relié à au moins deux candidatures.
+        tests:
+          - not_null
+      - name: updated_at
+        description: Timestamp de dernière mise à jour propagé depuis click/apply.
+        tests:
+          - not_null

analytics/dbt/analytics/models/intermediate/stat_event/int_stat_event_click_metrics.sql

@@ -0,0 +1,98 @@
+{{ config(
+  materialized = 'incremental',
+  unique_key = 'click_stat_event_id',
+  incremental_strategy = 'delete+insert',
+  on_schema_change = 'sync_all_columns',
+  post_hook = [
+    'create index if not exists "int_stat_event_click_metrics_mission_id_idx" on {{ this }} (mission_id)',
+    'create index if not exists "int_stat_event_click_metrics_from_publisher_id_idx" on {{ this }} (from_publisher_id)',
+    'create index if not exists "int_stat_event_click_metrics_click_created_at_idx" on {{ this }} (click_created_at)',
+    'create index if not exists "int_stat_event_click_metrics_click_stat_event_id_idx" on {{ this }} (click_stat_event_id)'
+  ]
+) }}
+
+with last_run as (
+  {% if is_incremental() %}
+    select coalesce(max(updated_at), '1900-01-01'::timestamp) as last_updated_at
+    from {{ this }}
+  {% else %}
+    select '1900-01-01'::timestamp as last_updated_at
+  {% endif %}
+),
+
+affected_clicks as (
+  select distinct e.stat_event_id as click_stat_event_id
+  from {{ ref('int_stat_event_global') }} as e
+  where
+    e.type = 'click'
+    and e.mission_id is not null
+    and coalesce(e.updated_at, e.created_at)
+    >= (select lr.last_updated_at from last_run as lr)
+
+  union
+
+  select distinct a.click_id as click_stat_event_id
+  from {{ ref('int_stat_event_global') }} as a
+  where
+    a.type = 'apply'
+    and a.click_id <> ''
+    and coalesce(a.updated_at, a.created_at)
+    >= (select lr.last_updated_at from last_run as lr)
+),
+
+clicks as (
+  select
+    e.stat_event_id as click_stat_event_id,
+    e.mission_id,
+    e.from_publisher_id,
+    e.source,
+    e.source_id,
+    e.created_at as click_created_at,
+    e.updated_at as click_updated_at
+  from {{ ref('int_stat_event_global') }} as e
+  where
+    e.type = 'click'
+    and e.mission_id is not null
+    {% if is_incremental() %}
+      and e.stat_event_id in (
+        select ac.click_stat_event_id from affected_clicks as ac
+      )
+    {% endif %}
+),
+
+applies_by_click as (
+  select
+    a.click_id as click_stat_event_id,
+    min(a.created_at) as first_apply_at,
+    count(*) as apply_count,
+    max(coalesce(a.updated_at, a.created_at)) as apply_updated_at
+  from {{ ref('int_stat_event_global') }} as a
+  where
+    a.type = 'apply'
+    and a.click_id <> ''
+    {% if is_incremental() %}
+      and a.click_id in (
+        select ac.click_stat_event_id from affected_clicks as ac
+      )
+    {% endif %}
+  group by 1
+)
+
+select
+  c.click_stat_event_id,
+  c.mission_id,
+  c.from_publisher_id,
+  c.source,
+  c.source_id,
+  c.click_created_at,
+  abc.first_apply_at,
+  coalesce(abc.apply_count, 0) as apply_count,
+  (coalesce(abc.apply_count, 0) >= 1) as has_apply,
+  (coalesce(abc.apply_count, 0) >= 2) as has_multi_apply,
+  greatest(
+    coalesce(c.click_updated_at, c.click_created_at),
+    coalesce(abc.apply_updated_at, '1900-01-01'::timestamp)
+  ) as updated_at
+from clicks as c
+left join applies_by_click as abc
+  on c.click_stat_event_id = abc.click_stat_event_id