complete app

dbt_nba/models/marts/facts/fct_game_results.sql

@@ -0,0 +1,50 @@
+{{
+    config(
+        materialized='incremental',
+        schema='marts',
+        unique_key='game_key',
+        tags=["fact"]
+    )
+}}
+
+WITH games_enriched AS (
+    SELECT *
+    FROM {{ ref('int_games_enriched') }}
+    {% if is_incremental() %}
+    WHERE game_date >= (SELECT MAX(game_date) FROM {{ this }}) - INTERVAL '30 days'
+    {% endif %}
+)
+
+SELECT
+    {{ dbt_utils.generate_surrogate_key(['ge.game_id']) }} AS game_key,
+    d.date_key,
+    s.season_key,
+    a.arena_key,
+    home_team.team_key AS home_team_key,
+    visitor_team.team_key AS visitor_team_key,
+    winning_team.team_key AS winning_team_key,
+    ge.game_id,
+    ge.home_points,
+    ge.visitor_points,
+    ge.point_differential,
+    ge.total_points,
+    ge.is_playoff,
+    ge.is_overtime,
+    CASE WHEN ge.home_points > ge.visitor_points THEN TRUE ELSE FALSE END AS is_home_team_winner,
+    ge.home_net_rating,
+    ge.visitor_net_rating,
+    ge.matchup_pace,
+    CASE
+        WHEN ge.point_differential <= 5 THEN 'Clutch Game'
+        WHEN ge.point_differential <= 10 THEN 'Competitive'
+        WHEN ge.point_differential <= 20 THEN 'Decisive'
+        ELSE 'Blowout'
+    END AS game_competitiveness_tier,
+    CAST(ge.game_date AS DATE) AS game_date
+FROM games_enriched AS ge
+LEFT JOIN {{ ref('dim_dates') }} AS d ON CAST(ge.game_date AS DATE) = d.full_date
+LEFT JOIN {{ ref('dim_seasons') }} AS s ON ge.season_start_year = s.season_start_year
+LEFT JOIN {{ ref('dim_arenas') }} AS a ON ge.arena = a.arena_name AND ge.arena_city = a.arena_city
+LEFT JOIN {{ ref('dim_teams') }} AS home_team ON ge.home_team = home_team.team_abbr
+LEFT JOIN {{ ref('dim_teams') }} AS visitor_team ON ge.visitor_team = visitor_team.team_abbr
+LEFT JOIN {{ ref('dim_teams') }} AS winning_team ON ge.winning_team = winning_team.team_abbr

dbt_nba/models/marts/facts/fct_player_game_shooting.sql

@@ -0,0 +1,175 @@
+{{
+    config(
+        materialized='incremental',
+        schema='marts',
+        unique_key='player_game_shooting_key',
+        tags=["fact"]
+    )
+}}
+
+WITH shots AS (
+    SELECT *
+    FROM {{ ref('int_player_shots_enriched') }}
+    WHERE has_game_match = TRUE
+    {% if is_incremental() %}
+      AND game_date >= (SELECT MAX(game_date) FROM {{ this }}) - INTERVAL '30 days'
+    {% endif %}
+),
+
+game_agg AS (
+    SELECT
+        game_id, player_id, team, opponent,
+        MIN(game_date) AS game_date,
+        MIN(season_start_year) AS season_start_year,
+        BOOL_OR(is_playoff) AS is_playoff,
+        MIN(team_location) AS team_location,
+        MIN(game_result) AS game_result,
+
+        SUM(points_generated) AS total_points,
+        COUNT(*) AS total_shot_attempts,
+        SUM(shot_made_flag) AS total_shots_made,
+
+        COUNT(*) FILTER (WHERE shot_source = 'shot_chart') AS fg_attempts,
+        SUM(shot_made_flag) FILTER (WHERE shot_source = 'shot_chart') AS fg_makes,
+        CASE WHEN COUNT(*) FILTER (WHERE shot_source = 'shot_chart') > 0
+            THEN SUM(shot_made_flag) FILTER (WHERE shot_source = 'shot_chart')::DECIMAL
+                 / COUNT(*) FILTER (WHERE shot_source = 'shot_chart')::DECIMAL
+            ELSE 0
+        END AS fg_pct,
+        SUM(points_generated) FILTER (WHERE shot_source = 'shot_chart') AS fg_points,
+
+        COUNT(*) FILTER (WHERE shot_source = 'box_score_ft') AS ft_attempts,
+        SUM(shot_made_flag) FILTER (WHERE shot_source = 'box_score_ft') AS ft_makes,
+        CASE WHEN COUNT(*) FILTER (WHERE shot_source = 'box_score_ft') > 0
+            THEN SUM(shot_made_flag) FILTER (WHERE shot_source = 'box_score_ft')::DECIMAL
+                 / COUNT(*) FILTER (WHERE shot_source = 'box_score_ft')::DECIMAL
+            ELSE 0
+        END AS ft_pct,
+        SUM(points_generated) FILTER (WHERE shot_source = 'box_score_ft') AS ft_points,
+
+        COUNT(*) FILTER (WHERE is_three_pointer = FALSE AND shot_source = 'shot_chart') AS two_point_attempts,
+        SUM(shot_made_flag) FILTER (WHERE is_three_pointer = FALSE AND shot_source = 'shot_chart') AS two_point_makes,
+        CASE WHEN COUNT(*) FILTER (WHERE is_three_pointer = FALSE AND shot_source = 'shot_chart') > 0
+            THEN SUM(shot_made_flag) FILTER (WHERE is_three_pointer = FALSE AND shot_source = 'shot_chart')::DECIMAL
+                 / COUNT(*) FILTER (WHERE is_three_pointer = FALSE AND shot_source = 'shot_chart')::DECIMAL
+            ELSE 0
+        END AS two_point_fg_pct,
+
+        COUNT(*) FILTER (WHERE is_three_pointer = TRUE) AS three_point_attempts,
+        SUM(shot_made_flag) FILTER (WHERE is_three_pointer = TRUE) AS three_point_makes,
+        CASE WHEN COUNT(*) FILTER (WHERE is_three_pointer = TRUE) > 0
+            THEN SUM(shot_made_flag) FILTER (WHERE is_three_pointer = TRUE)::DECIMAL
+                 / COUNT(*) FILTER (WHERE is_three_pointer = TRUE)::DECIMAL
+            ELSE 0
+        END AS three_point_fg_pct,
+
+        COUNT(*) FILTER (WHERE shot_distance_zone = 'At Rim (0-3 ft)') AS at_rim_attempts,
+        SUM(shot_made_flag) FILTER (WHERE shot_distance_zone = 'At Rim (0-3 ft)') AS at_rim_makes,
+        CASE WHEN COUNT(*) FILTER (WHERE shot_distance_zone = 'At Rim (0-3 ft)') > 0
+            THEN SUM(shot_made_flag) FILTER (WHERE shot_distance_zone = 'At Rim (0-3 ft)')::DECIMAL
+                 / COUNT(*) FILTER (WHERE shot_distance_zone = 'At Rim (0-3 ft)')::DECIMAL
+            ELSE 0
+        END AS at_rim_fg_pct,
+
+        COUNT(*) FILTER (WHERE is_clutch_shot = TRUE) AS clutch_fg_attempts,
+        SUM(shot_made_flag) FILTER (WHERE is_clutch_shot = TRUE) AS clutch_fg_makes,
+        CASE WHEN COUNT(*) FILTER (WHERE is_clutch_shot = TRUE) > 0
+            THEN SUM(shot_made_flag) FILTER (WHERE is_clutch_shot = TRUE)::DECIMAL
+                 / COUNT(*) FILTER (WHERE is_clutch_shot = TRUE)::DECIMAL
+            ELSE 0
+        END AS clutch_fg_pct,
+
+        CASE WHEN COUNT(*) FILTER (WHERE shot_source = 'shot_chart') > 0
+            THEN COUNT(*) FILTER (WHERE is_three_pointer = TRUE)::DECIMAL
+                 / COUNT(*) FILTER (WHERE shot_source = 'shot_chart')::DECIMAL
+            ELSE 0
+        END AS three_point_rate,
+
+        CASE WHEN COUNT(*) FILTER (WHERE shot_source = 'shot_chart') > 0
+            THEN COUNT(*) FILTER (WHERE shot_distance_zone = 'At Rim (0-3 ft)')::DECIMAL
+                 / COUNT(*) FILTER (WHERE shot_source = 'shot_chart')::DECIMAL
+            ELSE 0
+        END AS at_rim_rate,
+
+        CASE WHEN COUNT(*) FILTER (WHERE shot_source = 'shot_chart') > 0
+            THEN COUNT(*) FILTER (WHERE shot_distance_zone IN ('Mid Range (11-16 ft)', 'Long Mid Range (17-23 ft)'))::DECIMAL
+                 / COUNT(*) FILTER (WHERE shot_source = 'shot_chart')::DECIMAL
+            ELSE 0
+        END AS mid_range_rate,
+
+        AVG(distance_ft) FILTER (WHERE shot_source = 'shot_chart') AS avg_fg_distance_ft,
+
+        CASE WHEN (
+                COUNT(*) FILTER (WHERE shot_source = 'shot_chart')
+                + 0.44 * COUNT(*) FILTER (WHERE shot_source = 'box_score_ft')
+            ) > 0
+            THEN SUM(points_generated)::DECIMAL / (
+                2.0 * (
+                    COUNT(*) FILTER (WHERE shot_source = 'shot_chart')
+                    + 0.44 * COUNT(*) FILTER (WHERE shot_source = 'box_score_ft')
+                )
+            )
+            ELSE 0
+        END AS true_shooting_pct
+
+    FROM shots
+    GROUP BY game_id, player_id, team, opponent
+),
+
+final AS (
+    SELECT
+        {{ dbt_utils.generate_surrogate_key(['ga.game_id', 'ga.player_id']) }} AS player_game_shooting_key,
+        p.player_key,
+        t.team_key,
+        opp_t.team_key AS opponent_key,
+        d.date_key,
+        s.season_key,
+        ga.game_id,
+        ga.is_playoff,
+        ga.team_location,
+        ga.game_result,
+        ga.total_points,
+        ga.total_shot_attempts,
+        ga.total_shots_made,
+        ga.fg_attempts,
+        ga.fg_makes,
+        ga.fg_pct,
+        ga.fg_points,
+        ga.ft_attempts,
+        ga.ft_makes,
+        ga.ft_pct,
+        ga.ft_points,
+        ga.true_shooting_pct,
+        ga.two_point_attempts,
+        ga.two_point_makes,
+        ga.two_point_fg_pct,
+        ga.three_point_attempts,
+        ga.three_point_makes,
+        ga.three_point_fg_pct,
+        ga.at_rim_attempts,
+        ga.at_rim_makes,
+        ga.at_rim_fg_pct,
+        ga.clutch_fg_attempts,
+        ga.clutch_fg_makes,
+        ga.clutch_fg_pct,
+        ga.three_point_rate,
+        ga.at_rim_rate,
+        ga.mid_range_rate,
+        ga.avg_fg_distance_ft,
+        CASE
+            WHEN ga.three_point_rate >= 0.50 THEN 'Perimeter Heavy'
+            WHEN ga.at_rim_rate >= 0.50 THEN 'Rim Attacker'
+            WHEN ga.mid_range_rate >= 0.40 THEN 'Mid Range Heavy'
+            WHEN ga.three_point_rate >= 0.35 AND ga.at_rim_rate >= 0.30 THEN 'Modern (Rim & Three)'
+            ELSE 'Balanced'
+        END AS shot_profile_type,
+        CAST(ga.game_date AS DATE) AS game_date
+    FROM game_agg AS ga
+    LEFT JOIN {{ ref('dim_players') }} AS p ON ga.player_id = p.player_id
+    LEFT JOIN {{ ref('dim_teams') }} AS t ON ga.team = t.team_abbr
+    LEFT JOIN {{ ref('dim_teams') }} AS opp_t ON ga.opponent = opp_t.team_abbr
+    LEFT JOIN {{ ref('dim_dates') }} AS d ON CAST(ga.game_date AS DATE) = d.full_date
+    LEFT JOIN {{ ref('dim_seasons') }} AS s ON ga.season_start_year = s.season_start_year
+)
+
+SELECT * FROM final

dbt_nba/models/marts/facts/fct_player_game_stats.sql

@@ -0,0 +1,71 @@
+{{
+    config(
+        materialized='incremental',
+        schema='marts',
+        unique_key='player_game_key',
+        tags=["fact"]
+    )
+}}
+
+WITH player_performance AS (
+    SELECT * FROM {{ ref('int_player_performance') }}
+),
+
+game_details AS (
+    SELECT game_id, arena AS arena_name, arena_city
+    FROM {{ ref('int_games_enriched') }}
+),
+
+final AS (
+    SELECT
+        {{ dbt_utils.generate_surrogate_key(['pp.game_id', 'pp.player_id']) }} AS player_game_key,
+        p.player_key,
+        t.team_key,
+        d.date_key,
+        s.season_key,
+        a.arena_key,
+        arch.archetype_key,
+        pp.game_id,
+        pp.minutes_played,
+        pp.points,
+        pp.assists,
+        pp.total_rebounds,
+        pp.steals,
+        pp.blocks,
+        pp.turnovers,
+        pp.plus_minus,
+        pp.net_rating,
+        pp.box_plus_minus,
+        pp.field_goals_made,
+        pp.field_goals_attempted,
+        pp.three_pointers_made,
+        pp.three_pointers_attempted,
+        pp.true_shooting_pct,
+        pp.effective_fg_pct,
+        pp.usage_pct,
+        pp.offensive_rating,
+        pp.defensive_rating,
+        CAST(pp.game_date AS DATE) AS game_date
+    FROM player_performance AS pp
+    LEFT JOIN game_details AS gd ON pp.game_id = gd.game_id
+    LEFT JOIN {{ ref('dim_player_game_archetypes') }} AS arch
+        ON pp.usage_tier = arch.usage_tier
+        AND pp.impact_tier = arch.impact_tier
+        AND pp.shooting_efficiency_tier = arch.shooting_efficiency_tier
+        AND pp.minutes_based_role = arch.minutes_based_role
+        AND pp.is_double_double = arch.is_double_double
+        AND pp.is_triple_double = arch.is_triple_double
+        AND pp.is_versatile = arch.is_versatile
+        AND pp.is_defensive_specialist = arch.is_defensive_specialist
+        AND pp.is_three_and_d = arch.is_three_and_d
+    LEFT JOIN {{ ref('dim_players') }} AS p ON pp.player_id = p.player_id
+    LEFT JOIN {{ ref('dim_teams') }} AS t ON pp.team = t.team_abbr
+    LEFT JOIN {{ ref('dim_dates') }} AS d ON CAST(pp.game_date AS DATE) = d.full_date
+    LEFT JOIN {{ ref('dim_seasons') }} AS s ON pp.season_start_year = s.season_start_year
+    LEFT JOIN {{ ref('dim_arenas') }} AS a ON gd.arena_name = a.arena_name AND gd.arena_city = a.arena_city
+    {% if is_incremental() %}
+    WHERE pp.game_date >= (SELECT MAX(game_date) FROM {{ this }}) - INTERVAL '30 days'
+    {% endif %}
+)
+
+SELECT * FROM final

dbt_nba/models/marts/facts/fct_player_shots.sql

@@ -0,0 +1,62 @@
+{{
+    config(
+        materialized='incremental',
+        schema='marts',
+        unique_key='shot_key',
+        tags=["fact"]
+    )
+}}
+
+WITH shots_enriched AS (
+    SELECT *
+    FROM {{ ref('int_player_shots_enriched') }}
+    WHERE has_game_match = TRUE
+    {% if is_incremental() %}
+      AND game_date >= (SELECT MAX(game_date) FROM {{ this }}) - INTERVAL '30 days'
+    {% endif %}
+),
+
+final AS (
+    SELECT
+        {{ dbt_utils.generate_surrogate_key(['se.shot_id', 'se.shot_source']) }} AS shot_key,
+        p.player_key,
+        t.team_key,
+        opp_t.team_key AS opponent_key,
+        d.date_key,
+        s.season_key,
+        se.game_id,
+        se.shot_id,
+        se.shot_source,
+        se.is_playoff,
+        se.team_location,
+        se.game_result,
+        se.quarter_number,
+        se.is_overtime_shot,
+        se.seconds_remaining_in_quarter,
+        se.shot_x_coordinate,
+        se.shot_y_coordinate,
+        se.is_made,
+        se.shot_made_flag,
+        se.shot_missed_flag,
+        se.shot_type_raw AS shot_type,
+        se.shot_point_value,
+        se.is_three_pointer,
+        se.is_free_throw,
+        se.distance_ft,
+        se.shot_distance_zone,
+        se.points_generated,
+        se.team_had_lead,
+        se.team_score_at_shot,
+        se.opponent_score_at_shot,
+        se.score_margin_at_shot,
+        se.is_clutch_shot,
+        CAST(se.game_date AS DATE) AS game_date
+    FROM shots_enriched AS se
+    LEFT JOIN {{ ref('dim_players') }} AS p ON se.player_id = p.player_id
+    LEFT JOIN {{ ref('dim_teams') }} AS t ON se.team = t.team_abbr
+    LEFT JOIN {{ ref('dim_teams') }} AS opp_t ON se.opponent = opp_t.team_abbr
+    LEFT JOIN {{ ref('dim_dates') }} AS d ON CAST(se.game_date AS DATE) = d.full_date
+    LEFT JOIN {{ ref('dim_seasons') }} AS s ON se.season_start_year = s.season_start_year
+)
+
+SELECT * FROM final

dbt_nba/models/marts/facts/fct_quarter_scoring.sql

@@ -0,0 +1,57 @@
+{{
+    config(
+        materialized='incremental',
+        schema='marts',
+        unique_key='quarter_scoring_key',
+        tags=["fact"]
+    )
+}}
+
+WITH games AS (
+    SELECT game_id, game_date, season_start_year
+    FROM {{ ref('int_games_enriched') }}
+    {% if is_incremental() %}
+    WHERE game_date >= (SELECT MAX(game_date) FROM {{ this }}) - INTERVAL '30 days'
+    {% endif %}
+),
+
+unpivoted_scores AS (
+    {{ dbt_utils.unpivot(
+        relation=ref('stg_line_scores'),
+        cast_to='INTEGER',
+        exclude=['game_id', 'team', 'total_points', 'first_half_points', 'second_half_points', 'regulation_points', 'overtime_points', 'had_ot1', 'had_ot2', 'had_ot3', 'q2_momentum', 'q3_momentum', 'q4_momentum', 'max_quarter_score', 'min_quarter_score', 'best_quarter', 'created_at', 'updated_at', 'dbt_loaded_at'],
+        field_name='period_name',
+        value_name='points_scored'
+    ) }}
+    WHERE game_id IN (SELECT game_id FROM games)
+),
+
+final AS (
+    SELECT
+        {{ dbt_utils.generate_surrogate_key(['team_scores.game_id', 'team_map.team_abbr', 'team_scores.period_name']) }} AS quarter_scoring_key,
+        d.date_key,
+        s.season_key,
+        t.team_key,
+        opp.team_key AS opponent_key,
+        team_scores.game_id,
+        REPLACE(UPPER(team_scores.period_name), '_POINTS', '') AS period,
+        team_scores.points_scored,
+        opponent_scores.points_scored AS opponent_points_scored,
+        (team_scores.points_scored - opponent_scores.points_scored) AS period_point_differential,
+        CAST(g.game_date AS DATE) AS game_date
+    FROM unpivoted_scores AS team_scores
+    INNER JOIN unpivoted_scores AS opponent_scores
+        ON team_scores.game_id = opponent_scores.game_id
+        AND team_scores.period_name = opponent_scores.period_name
+        AND team_scores.team != opponent_scores.team
+    INNER JOIN games AS g ON team_scores.game_id = g.game_id
+    LEFT JOIN {{ ref('team_maps') }} AS team_map ON team_scores.team = team_map.team_abbr
+    LEFT JOIN {{ ref('team_maps') }} AS opponent_map ON opponent_scores.team = opponent_map.team_abbr
+    LEFT JOIN {{ ref('dim_teams') }} AS t ON team_map.team_abbr = t.team_abbr
+    LEFT JOIN {{ ref('dim_teams') }} AS opp ON opponent_map.team_abbr = opp.team_abbr
+    LEFT JOIN {{ ref('dim_dates') }} AS d ON CAST(g.game_date AS DATE) = d.full_date
+    LEFT JOIN {{ ref('dim_seasons') }} AS s ON g.season_start_year = s.season_start_year
+    WHERE team_scores.points_scored IS NOT NULL
+)
+
+SELECT * FROM final

dbt_nba/models/marts/facts/fct_team_game_stats.sql

@@ -0,0 +1,54 @@
+{{
+    config(
+        materialized='incremental',
+        schema='marts',
+        tags=["fact"],
+        unique_key='team_game_key'
+    )
+}}
+
+WITH team_performance AS (
+    SELECT * FROM {{ ref('int_team_performance') }}
+),
+
+game_details AS (
+    SELECT
+        game_id, game_date, season_start_year,
+        home_team, visitor_team, arena AS arena_name, arena_city
+    FROM {{ ref('int_games_enriched') }}
+),
+
+final AS (
+    SELECT
+        {{ dbt_utils.generate_surrogate_key(['tp.game_id', 'tp.team']) }} AS team_game_key,
+        t.team_key,
+        d.date_key,
+        s.season_key,
+        a.arena_key,
+        opp.team_key AS opponent_key,
+        tp.game_id,
+        (tp.team = gd.home_team) AS is_home_team,
+        tp.points,
+        tp.offensive_rating,
+        tp.defensive_rating,
+        tp.net_rating,
+        tp.pace,
+        tp.effective_fg_pct,
+        tp.turnover_rate,
+        tp.offensive_tier,
+        tp.defensive_tier,
+        CAST(gd.game_date AS DATE) AS game_date
+    FROM team_performance AS tp
+    LEFT JOIN game_details AS gd ON tp.game_id = gd.game_id
+    LEFT JOIN {{ ref('dim_teams') }} AS t ON tp.team = t.team_abbr
+    LEFT JOIN {{ ref('dim_teams') }} AS opp
+        ON CASE WHEN tp.team = gd.home_team THEN gd.visitor_team ELSE gd.home_team END = opp.team_abbr
+    LEFT JOIN {{ ref('dim_dates') }} AS d ON CAST(gd.game_date AS DATE) = d.full_date
+    LEFT JOIN {{ ref('dim_seasons') }} AS s ON gd.season_start_year = s.season_start_year
+    LEFT JOIN {{ ref('dim_arenas') }} AS a ON gd.arena_name = a.arena_name AND gd.arena_city = a.arena_city
+    {% if is_incremental() %}
+    WHERE gd.game_date >= (SELECT MAX(game_date) FROM {{ this }}) - INTERVAL '30 days'
+    {% endif %}
+)
+
+SELECT * FROM final