The objective of this project was to simulate regularized (LASSO and Elastic-net) linear and logistic regression models for feature selection in order to verify results about the accuracy of these feature selection methods, as presented in the following paper:
Florentina Bunea. 2008. Honest variable selection in linear and logistic regression models
via l1 and l1 + l2 penalization. Electronic Journal of Statistics, 2:1153โ1194
For the project, we did the following:
- Implemented linear and logistic regression feature selection models with LASSO and Elastic-net regularization
- Designed and executed feature selection simulations for varying sample sizes and noise levels
- Created visualizations of model performance indicators for each simulation run
The code for the project can be found in Regularized_Subset_Selection.ipynb, and the final report can be viewed in Regularized_Subset_Selection_Report.pdf.