← back to experiments

§ 2.1 — EXP-04 / Sports Analytics

nba-predictive-modeling

An active machine learning system for forecasting Golden State Warriors game outcomes.

Project overview

This project aims to:

  • Analyze historical Warriors game data using feature engineering.
  • Build ML models to predict game outcomes.
  • Implement an incremental learning system that updates with each new game.
  • Make predictions for upcoming games using current season data.

Current implementation

  • Data pipeline over historical and current season game data.
  • Feature engineering for rolling statistics, streaks, rest days, and matchup context.
  • Multiple model tracks for classification (win/loss) and regression (point differential).
  • Prediction output generation for upcoming games.

Next step

Continue improving feature variation and incremental update stability to tighten calibration and weekly prediction consistency.