Webinar: Malgorzata Bogdan (Lund) February 9

The first webinar of the spring 2024 will be given by Malgorzata Bogdan (Lund).

When: Friday February 9, 11.00 – 11.45

Where: Zoom link: https://lu-se.zoom.us/j/61931718764?pwd=YURHL01WamJPOUhhWmY3S3NTR0VrUT09

Meeting ID: 619 3171 8764
Password: 047554

Titel: On the properties of LASSO and SLOPE.

Abstract: LASSO and SLOPE are two regularization methods that can be used to reduce the dimensionality of overparametrized regression models. LASSO eliminates certain regression coefficients, while SLOPE  additionally allows for some coefficients to be exactly equal. However, both methods require the application of large penalty coefficients to achieve these model selection properties, which can result in substantial bias and deterioration of estimation and model selection properties. In this presentation, we will discuss several new theoretical findings that shed light on the properties of LASSO and SLOPE, as well as their preferred thresholded and adaptive versions. If time permits, we will also showcase some applications of SLOPE in the context of portfolio optimization.