Reduction Rank Regression

Reduced rank regression is a model-based approach to simultaneous dimension reduction and regression, where rather than including each covariate separately in a model, linear combinations of the covariates (which often have some form of interpretation to them) are used instead. This project will review the benefits of this approach, potential extensions of this approach, and potentially explore key questions such as choosing the rank and how to perform inference.

This project will be co-supervised with Dr Francis Hui.