Date of Completion
Spring 6-12-2019
Thesis Advisor(s)
Nehemy Lim
Honors Major
Statistics
Abstract
This paper presents the FOS algorithm as first outlined by Lim and Lederer in 2016, and describes its intuition. FOS is an algorithm that efficiently traverses the L1 regularization path of the LASSO regression. This paper also presents a novel implementation of the FOS algorithm for the Group LASSO problem, and compares this algorithm against a state-of-the-art regression package. These algorithms both have potential for significant impact in biomedical research regarding the analysis of gene expression data.
Recommended Citation
Pandit, Saket, "Fast Optimization Algorithm for Linear Regression Models with Sparsity-Inducing Penalties" (2019). Honors Scholar Theses. 632.
https://digitalcommons.lib.uconn.edu/srhonors_theses/632