Title
A Robust Estimation Framework for Spatiotemporal Epidemic Models
Abstract
Producing accurate estimates and correctly understanding the spatiotemporal effects are crucial aspects of epidemic modeling. However, these processes can be frequently impacted by the presence of outliers in the data, which may bias estimation. In this work, we propose a robust estimation framework for spatiotemporal models by introducing a slack variable in a generalized geoadditive model to account for the effects of outliers. The model estimation is achieved via proximal methods. Through simulations, our model demonstrates the capability of capturing the underlying true spatial effects, correctly identifying outliers, and providing accurate magnitude estimates in various cases. Our proposed method fills a gap in the spatiotemporal modeling literature, offering a novel statistical framework for robust estimation, and having the potential for applications to a broader range of spatiotemporal fields beyond epidemic scenarios.
Bio
I am a second-year PhD student in Statistics at Rice Unveristy, mentoring by Dr. Eric Chi. Before joining Rice, I graduated from UC - Santa Barbara in 2022 with a major of Financial Math and Statistics. I am currently working on projects about robust estimation, survival clustering, and fast MCMC. And I am broadly interested in numerical optimization, machine learning, high-dimensional statistics, and their applications.