guitavares8/Football_ai_Predictor

A Machine Learning system to produce quantitative predictions for Portuguese Primeira Liga matches. It targets sports analysts, data scientists and bettors who need probabilistic forecasts rather than simple averages. The pipeline combines historical results, an Elo ranking, advanced form features (streaks, rolling Expected Goals) and an exponential time-decay to prioritize recent form. Key features: - Three real-time models served by a Streamlit app: XGBClassifier for 1X2 probabilities, XGBRegressor for total-goals expectation (combined with a Poisson distribution to derive exact over/under odds), and an XGBClassifier for Both Teams To Score (BTTS). - Conversion of probabilities into double-chance coverages. - Persisted models (.pkl), preprocessing and evaluation notebooks, data ingestion automation. Tech stack: Python, pandas/numpy, scikit-learn, XGBoost, Streamlit. Repository separates data, trained models, notebooks and the app.

AI-related keywords found in this repo's description, topics, or README summary — grouped by category. Each badge links to the corresponding ranking detail page.