Player skill snapshot

models/marts/player_skill_snapshot.sql

@@ -0,0 +1,75 @@
+{{ config(location='data_export/player_skill_snapshot.parquet') }}
+
+WITH
+  matches AS (
+    SELECT
+      match_id,
+      start_time,
+      game_type,
+      lower(game_type) AS game_type_l
+    FROM {{ ref("matches") }}
+    WHERE is_ranked = true
+  ),
+  players AS (
+    SELECT
+      user_id,
+      name,
+      country AS country_code
+    FROM {{ ref("players") }}
+  ),
+  joined AS (
+    SELECT
+      m.start_time,
+      mp.user_id,
+      p.name,
+      p.country_code,
+      mp.new_skill,
+      mp.new_uncertainty,
+      CASE
+        WHEN m.game_type_l LIKE '%duel%' THEN 'duel'
+        WHEN m.game_type_l LIKE '%ffa%' THEN 'ffa'
+        WHEN m.game_type_l LIKE '%large%' THEN 'large'
+        WHEN m.game_type_l LIKE '%small%' THEN 'small'
+      END AS game_type
+    FROM {{ ref("match_players") }} AS mp
+    INNER JOIN matches AS m ON mp.match_id = m.match_id
+    INNER JOIN players AS p ON mp.user_id = p.user_id
+  ),
+  aggregated AS (
+    SELECT
+      user_id,
+      arg_max(new_skill, start_time) FILTER (WHERE game_type = 'duel') AS duel_skill,
+      arg_max(new_uncertainty, start_time) FILTER (WHERE game_type = 'duel') AS duel_skill_un,
+      arg_max(new_skill, start_time) FILTER (WHERE game_type = 'ffa') AS ffa_skill,
+      arg_max(new_uncertainty, start_time) FILTER (WHERE game_type = 'ffa') AS ffa_skill_un,
+      arg_max(new_skill, start_time) FILTER (WHERE game_type = 'large') AS team_skill,
+      arg_max(new_uncertainty, start_time) FILTER (WHERE game_type = 'large') AS team_skill_un,
+      arg_max(start_time, start_time) FILTER (WHERE game_type = 'duel') AS last_duel,
+      arg_max(start_time, start_time) FILTER (WHERE game_type = 'ffa') AS last_ffa,
+      arg_max(start_time, start_time) FILTER (WHERE game_type = 'large') AS last_team,
+      arg_max(start_time, start_time) FILTER (WHERE game_type = 'small') AS last_small_team,
+      arg_max(new_skill, start_time) FILTER (WHERE game_type = 'small') AS small_team_skill,
+      arg_max(new_uncertainty, start_time) FILTER (WHERE game_type = 'small') AS small_team_skill_un
+    FROM joined
+    WHERE game_type IS NOT null
+    GROUP BY user_id
+  )
+SELECT
+  a.user_id AS id,
+  p.name,
+  a.duel_skill,
+  a.duel_skill_un,
+  a.ffa_skill,
+  a.ffa_skill_un,
+  a.team_skill,
+  a.team_skill_un,
+  a.last_duel,
+  a.last_ffa,
+  a.last_team,
+  p.country_code,
+  a.last_small_team,
+  a.small_team_skill,
+  a.small_team_skill_un
+FROM aggregated AS a
+INNER JOIN players AS p ON a.user_id = p.user_id
+ORDER BY a.user_id

models/marts/schema.yml

@@ -123,8 +123,61 @@ models:
       - name: name
         data_type: string
         description: Current user selected name
+        data_tests:
           - unique
           - not_null
       - name: country
         data_type: string
         description: Country code of the player, e.g. PL, US, or NULL if hidden by player
+
+  - name: player_skill_snapshot
+    description: Latest OpenSkill snapshot per player and game type.
+    columns:
+      - name: id
+        data_type: bigint
+        description: Teiserver account id of the user
+        data_tests:
+          - unique
+          - not_null
+      - name: name
+        data_type: string
+        description: Current user selected name
+      - name: duel_skill
+        data_type: float
+        description: Latest duel skill (mu)
+      - name: duel_skill_un
+        data_type: float
+        description: Latest duel uncertainty (sigma)
+      - name: ffa_skill
+        data_type: float
+        description: Latest FFA skill (mu)
+      - name: ffa_skill_un
+        data_type: float
+        description: Latest FFA uncertainty (sigma)
+      - name: team_skill
+        data_type: float
+        description: Latest large team skill (mu)
+      - name: team_skill_un
+        data_type: float
+        description: Latest large team uncertainty (sigma)
+      - name: last_duel
+        data_type: timestamp
+        description: Time of latest duel match for this user
+      - name: last_ffa
+        data_type: timestamp
+        description: Time of latest FFA match for this user
+      - name: last_team
+        data_type: timestamp
+        description: Time of latest large team match for this user
+      - name: country_code
+        data_type: string
+        description: Country code of the player, e.g. PL, US, or NULL if hidden by player
+      - name: last_small_team
+        data_type: timestamp
+        description: Time of latest small team match for this user
+      - name: small_team_skill
+        data_type: float
+        description: Latest small team skill (mu)
+      - name: small_team_skill_un
+        data_type: float
+        description: Latest small team uncertainty (sigma)

scripts/build_more_formats.sql

@@ -3,3 +3,4 @@
 COPY 'data_export/matches.parquet' TO 'data_export/matches.csv.gz';
 COPY 'data_export/match_players.parquet' TO 'data_export/match_players.csv.gz';
 COPY 'data_export/players.parquet' TO 'data_export/players.csv.gz';
+COPY 'data_export/player_skill_snapshot.parquet' TO 'data_export/player_skill_snapshot.csv';