SebastianAndreu/24679_NFL_WR_Dataset_2015

license: mit task_categories:

• —tabular-regression
• —tabular-classification tags:
• —nfl
• —football
• —sports-analytics
• —wide-receiver
• —play-by-play size_categories:
• —1K<n<10K ---

NFL Wide Receiver Performance Dataset (2015)

Dataset Description

This dataset contains comprehensive wide receiver performance statistics derived from NFL play-by-play data for the 2015. It includes game-level metrics, situational targeting patterns, defensive adjustments, and advanced efficiency calculations.

Dataset Summary

• —Season: 2015
• —Records: ~4.31k player-game observations
• —Features: 100+ columns including:
• —Core statistics (targets, receptions, yards, touchdowns)
• —Quarter-by-quarter breakdowns
• —Win probability bucketed performance
• —Defensive strength adjustments
• —Situational metrics (red zone, high leverage, down-and-distance)
• —Team share metrics (target share, air yards share, WOPR)
• —Efficiency metrics (aDOT, yards per target, catch rate)
• —Weather and venue conditions

Supported Tasks

• —Receiving Yards Prediction: Predict receiving yards for upcoming games
• —Target Share Analysis: Model player opportunity distribution
• —Performance Forecasting: Project future player performance
• —Matchup Analysis: Evaluate player-defense matchups

Dataset Structure

Data Fields

Key Identifiers:

• —game_id: Unique game identifier
• —receiver_player_id: NFL GSIS player ID
• —receiver_player_name: Player display name
• —passer_player_id: Quarterback player ID
• —season: NFL season year
• —week: Week number

Core Statistics:

• —targets: Total pass attempts targeting the receiver
• —receptions: Completed receptions
• —receiving_yards: Total receiving yards
• —tds: Receiving touchdowns
• —air_yards: Total air yards on targets
• —yac: Yards after catch

Quarter Breakdowns:

• —yards_Q1, yards_Q2, yards_Q3, yards_Q4: Yards by quarter
• —receptions_Q1-4: Receptions by quarter
• —targets_Q1-4: Targets by quarter

Win Probability Buckets:

• —yards_wp_<25, yards_wp_25_45, etc.: Performance in different game situations
• —Similar breakdowns for receptions and targets

Share Metrics:

• —target_share: Player's share of team targets
• —air_yards_share: Player's share of team air yards
• —yard_share: Player's share of team receiving yards
• —reception_share: Player's share of team receptions
• —wopr: Weighted Opportunity Rating (0.7 × targetshare + 0.3 × airyards_share)

Efficiency Metrics:

• —aDOT: Average depth of target
• —yards_per_target: Receiving yards per target
• —catch_rate: Reception percentage
• —yac_per_rec: Yards after catch per reception
• —explosive_rec_pct: Percentage of receptions ≥15 yards
• —first_down_pct: Percentage of receptions resulting in first downs

Defensive Adjustments:

• —def_targets_dev: Defense targets allowed vs league average
• —def_receptions_dev: Defense receptions allowed vs league average
• —def_yards_dev: Defense yards allowed vs league average
• —def_tds_dev: Defense TDs allowed vs league average
• —def_epa_dev: Defense EPA allowed vs league average
• —adj_epa: Defense-adjusted Expected Points Added
• —adj_epa_per_target: Defense-adjusted EPA per target

Situational Metrics:

• —red_zone_targets: Targets inside the 20-yard line
• —end_zone_targets: Targets in the end zone
• —third_down_targets: Targets on 3rd down
• —fourth_down_targets: Targets on 4th down
• —high_leverage_targets: Targets in high-leverage situations (WP < 0.25 or > 0.75)
• —red_zone_share: Player's share of team red zone targets
• —third_down_share: Player's share of team 3rd down targets

Game Context:

• —posteam: Player's team (encoded 1-32)
• —defteam: Opposing defense (encoded 1-32)
• —home_team: Home team (encoded 1-32)
• —away_team: Away team (encoded 1-32)
• —home_flag: 0 if home, 1 if away
• —pregame_spread: Betting line point spread
• —pregame_total: Betting line total points
• —avg_score_diff: Average score differential when targeted
• —avg_quarter: Average quarter when targeted
• —trailing_pct: Percentage of targets while trailing
• —leading_pct: Percentage of targets while leading

Weather & Venue:

• —surface: Playing surface type (encoded 0-6)
• —is_dome: 1 if indoor, 0 if outdoor
• —is_rain: 1 if rainy conditions
• —is_snow: 1 if snowy conditions
• —is_clear: 1 if clear conditions
• —temp_f: Temperature in Fahrenheit
• —humidity_pct: Humidity percentage
• —wind_mph: Wind speed in miles per hour

QB Context:

• —qb_completions: Quarterback's completions that game
• —qb_attempts: Quarterback's attempts that game
• —qb_comp_pct: Quarterback's completion percentage
• —qb_air_yards: Quarterback's average air yards
• —qb_cpoe: Quarterback's completion percentage over expected

Advanced Metrics:

• —epa: Expected Points Added
• —wpa: Win Probability Added
• —success_rate: Percentage of successful plays (EPA > 0 or YPT > 0.5)
• —big_play_rate: Percentage of plays ≥20 yards
• —explosive_plays: Count of plays ≥20 yards
• —first_downs: First downs generated
• —consistency_score: meanadjepa / stdadjepa
• —inverse_volatility: 1 / stdadjepa
• —season_adj_epa_per_target: Season-level defense-adjusted EPA per target
• —wp_var: Variance in win probability across targets
• —target_share_std: Standard deviation of target share across games

Data Splits

This dataset does not include pre-defined splits. Users should create their own train/validation/test splits based on their use case:

• —Time-based split: Use early weeks for training, later weeks for validation/testing
• —Cross-validation: K-fold cross-validation across games
• —Season holdout: Train on this season, test on future seasons

Dataset Creation

Source Data

Raw play-by-play data sourced from nflverse, which aggregates official NFL data with additional features.

Data Processing

The dataset was created through two complementary processing pipelines:

1. 1.Pipeline A (Defensive Adjustments):
2. 2.Calculates defense-adjusted performance metrics
3. 3.Adds situational targeting patterns
4. 4.Includes QB context and team-level statistics
5. 5.Incorporates weather and venue conditions

1. 1.Pipeline B (Temporal & Situational):
2. 2.Generates quarter-by-quarter breakdowns
3. 3.Creates win probability bucketed statistics
4. 4.Computes team share metrics and WOPR
5. 5.Calculates season-level consistency metrics

Both pipelines were merged to create a comprehensive feature set.

Considerations for Using the Data

Social Impact

This dataset is intended for:

• —Sports analytics and research
• —Fantasy football decision-making
• —Educational purposes in machine learning and sports statistics

Not intended for:

• —Real-money gambling (use responsibly)
• —Player evaluation for contract negotiations
• —Any decision-making that could impact player careers

Discussion of Biases

• —Opportunity bias: Statistics heavily dependent on team offensive scheme and QB quality
• —Injury data: Dataset does not account for injuries that may affect performance
• —Sample size: Players with limited playing time have less reliable statistics
• —Game script: Performance metrics influenced by whether team is winning/losing
• —Strength of schedule: Not all defensive matchups are equal, though some adjustment is included

Limitations

• —Historical data only: Does not predict future performance definitively
• —Missing context: Does not include play design, route running, or other qualitative factors
• —Weather parsing: Temperature/wind/humidity may be missing or inaccurate for some games
• —Roster changes: Does not account for mid-season team changes or trades
• —Playoff games: May or may not include playoff data depending on the year

Additional Information

Acknowledgments

• —Data Source: nflverse/nflverse-data
• —AI Assistance: Code development assisted by Claude (Anthropic)
• —Course: CMU 24-679: Designing and Deploying AI/ML Systems