Details

This project presents a from-scratch implementation of LASSO (Least Absolute Shrinkage and Selection Operator) regression using the Homotopy Algorithm. Built entirely with NumPy, the model efficiently traces the full solution path as the regularization parameter decreases, encouraging sparse feature selection. It is designed to handle both small and collinear datasets, offering visualization of coefficient evolution and thorough test coverage. The project provides insights into the optimization process of LASSO without relying on external libraries like scikit-learn, and demonstrates strong reproducibility through test cases and visualizations.