Skip to main content

Panel studies and Granger causality in high dimensions

New technological advances have resulted in a rapid increase in generation of high-dimensional time series consisting of many variables. Examples include data from biological experiments, social networks, E-commerce and brain imaging. In this talk, I will focus primarily on panel studies, or longitudinal analysis, where multiple time series are observed over a relatively short period of time. I will describe penalized estimation methods for discovering interactions among variables using the framework of Granger causality. I will discuss methodological challenges, as well as opportunities for future research and collaboration.