How 333ian/football-ai-platform works
Football AI Prediction Platform serves match-outcome probabilities through a React dashboard and FastAPI, backed by PostgreSQL and trained XGBoost/Elo artifacts. The principal flow is team selection, prediction request, database team validation and model inference, then probability display. Separate ingestion scripts populate match/team records; model training reads historical results and writes artifacts. The README also describes a daily updater and batch forecasting. Some subsystem connections are documented rather than visible in sampled source, so those links are marked accordingly.
Main entry points: api.py REST API · database_setup.py PostgreSQL schema · train_model.py Training pipeline · App.jsx Match dashboard
Components 13
Match dashboard
- Match dashboard
frontend/src/App.jsx
Prediction service
- REST API
api.py - Batch forecasts
batch_predict.py
Match data
- PostgreSQL schema
database_setup.py - Tournament ingestion
mass_ingestion.py - Historical ingestion
historical_ingestion.py - Fixture ingestion
data-ingestion.py
Model lifecycle
- Training pipeline
train_model.py - Elo and features
train_model.py - Model artifacts
ml_models/saved_models - Daily updater
auto_updater.py
Other components
- Football user
- Sports data API
Connections 19
- Football user → Match dashboard: selects matchup
README.md - Match dashboard → REST API: requests teams
frontend/src/App.jsx - Match dashboard → REST API: submits prediction
frontend/src/App.jsx - REST API → PostgreSQL schema: queries teams
api.py - REST API → Model artifacts: loads model and Elo
api.py - REST API → Match dashboard: returns probabilities
frontend/src/App.jsx - Tournament ingestion → Sports data API: fetches tournaments
mass_ingestion.py - Historical ingestion → Sports data API: fetches results
historical_ingestion.py - Fixture ingestion → Sports data API: fetches fixtures
data-ingestion.py - Tournament ingestion → PostgreSQL schema: writes teams and matches
mass_ingestion.py - Historical ingestion → PostgreSQL schema: writes teams and matches
historical_ingestion.py - Fixture ingestion → PostgreSQL schema: writes teams and matches
data-ingestion.py - Training pipeline → PostgreSQL schema: reads finished matches
train_model.py - Training pipeline → Elo and features: calculates ratings and features
train_model.py - Training pipeline → Model artifacts: saves model and ratings
train_model.py - Daily updater → Training pipeline: re-trains model
README.md - Daily updater → Sports data API: fetches daily results
README.md - Batch forecasts → REST API: posts forecasts
batch_predict.py - Batch forecasts → PostgreSQL schema: queries scheduled fixtures
batch_predict.py