Streamlit
PythonScikit-Learn
Three.js
SHAP
Machine Learning
FIFA World Cup Match Predictor is an end-to-end machine learning system that predicts international football match outcomes (Win / Draw / Loss) using historical international results. The dataset is enriched with StatsBomb event data, Elo ratings, and player aggregates. The application opens with a stunning Three.js splash screen featuring 48 national flags orbiting a football, leading into an interactive dashboard. The dashboard offers calibrated matchday predictions, SHAP feature explanations for transparency, temporal validation & testing, and a custom matchup builder with an integrated xG simulation engine.
Features
- •Three.js Splash Screen: Features a 3D revolving carousel of 48 national flags orbiting a football, transitioning into the main dashboard.
- •Match Outcome Predictor: Calibrated probability estimates for custom matchups using 28,000+ training rows since 2000.
- •SHAP Explanations: Visualizes feature importances (Elo, form, Head-to-Head) to explain why the model favors a team.
- •xG Simulation Engine: Simulate goal outcomes and match scenarios based on granular shot and event metrics.
- •Daily Matchday Refresh: Scheduled data pipeline to update WC 2026 fixture picks and cache results.