Title: Multivariate Factorisable Expectile Regression with Application to fMRI Data





In this paper, we propose a multivariate expectile regression model for analyzing the tail events of large cross-sectional and spatial data, when the tail events show a latent factor structure. We show the computational benefit of our method, and we analyze the estimation risk under a finite iteration and finite sample setting, when the latent factors are exactly or approximately sparse. We apply our method on functional magnetic resonance imaging (fMRI) data generated from an experiment of making investment decisions, and show that the negative extreme blood oxygenation level dependent (BOLD) responses may be relevant to risk preferences.

Add to my calendar

Back to listing