Date Thesis Awarded


Access Type

Honors Thesis -- Access Restricted On-Campus Only

Degree Name

Bachelors of Science (BS)




Guannan Wang

Committee Members

Ross Iaci

Bin Ren


Advancements in information technology have enabled scientists to collect data of unprecedented size as well as complexity. Nowadays, high-dimensional data commonly arise in diverse fields as biology, engineering, health sciences, and economics. In this project, we consider both linear and non-parametric models with variable selection in the high-dimensional setting by assuming that only a small number of index coefficients influence the conditional mean of the response variable. Both the numerical results and the real data application demonstrate that the proposed approach selects the correct model with a high frequency and estimates the model coefficients accurately even for moderate sample size and ultra-high dimensionality.

On-Campus Access Only