NFL Spread Predictor
AI/MLAn integrated full-stack application predicting NFL spreads with live data, advanced ML techniques including XGBoost, and secure user authentication.
This comprehensive machine learning project represents a full-stack solution for predicting NFL game spreads using advanced data science techniques. I scraped and cleaned over 7 years of NFL data, processing more than 500,000 raw statistics into a structured dataset with 64,000+ features per row, including detailed team statistics, player performance metrics, and momentum indicators.
The core utilizes XGBoost with SMOTE oversampling and custom class weights, achieving 75% balanced accuracy against Vegas betting lines. I directly integrated the trained model into a Next.js frontend using custom hooks, providing real-time predictions with confidence scores and feature importance analysis. The technical stack spans the full data science pipeline from web scraping with Python/Pandas to modern web development with Next.js.