This repo contains implementations of univariate and multivariate regression in Machine Learning with polynomial and modified relu basis functions along with and without L2 regularization.
SOWC_combined_simple.csv
- Dataset for the implementation
data_utils.py
- Base code containing implementations for
- loading and cleaning the data
- z-normalizing (standardizing) features
- modified relu basis function
- computing design matrix for polynomial and relu basis functios using data
- estimating weights for learning with and without regularization
- evaluating the regressor
polynomial_regression.py
- Using base code to perform unregularized multivariate polynomial regressions of degree 1-6
polynomial_regression_1d.py
- Using base code to perform unregularized univariate polynomial regression of degree 3
visualize_1d.py
- To visualize the learned curve
relu_regression.py
- Using base code to perform unregularized multivariate relu regression
polynomial_regression_reg.py
- Using base code to perform L2 regularized multivariate polynomial regression of degree 2
- Identifying the best regularization constant using 10 fold cross-validation implementation
Folders inv and pinv contain visualizations.
For execution:
python3 filename.py