Mathematical Sciences Department PhD Dissertation Proposal Presentation - Lingli Yang, PhD Candidate (UH 420)

Wednesday, December 13, 2023
1:00 p.m. to 2:30 p.m.
Location
Floor/Room #
420
Preview

flyer

Mathematical Sciences Department

PhD Dissertation Proposal Presentation

 

Lingli Yang, PhD Candidate, Mathematical Sciences

Wednesday, December 13, 2023

1:00pm - 2:30pm

Unity Hall 420

Zoom Meeting ID: 952 6630 3718

https://wpi.zoom.us/j/95266303718

Title: Bayesian Predictive Inference with Stick-Breaking Processes and Survey Weights 

Abstract: We perform robust Bayesian inference by incorporating covariates, survey weights, and the stick-breaking process, seeking to overcome the inherent limitations of traditional analytical approaches. Our incorporation of covariates provides additional information about the subjects or units under study, and that of survey weights directs heightened attention towards observations with higher weights. The stick-breaking process, a fundamental component in Bayesian nonparametric models, enables a more flexible and robust modeling of data. Furthermore, we introduce a Bayesian predictive framework for binary and continuous study variables that makes use of probability survey samples that have been enhanced with auxiliary information. We compare different types of survey weights with unnormalized or normalized densities. Then, we propose the novel Bayesian nonparametric models based on the stick-breaking process and also incorporating survey weights. We describe an application on body mass index. Finally, we discuss the promising extensions of the robust Bayesian model with Stick-breaking process, one using a binary study variable and the other with the continuous study variable.

Dissertation Committee: 

Dr. Balgobin Nandram, WPI (Advisor)

Dr. Andrea Arnold, WPI

Dr. Adam Sales, WPI

Dr. Fangfang Wang, WPI

Dr. Jai Won Choi, Statistical Consultant, Meho Inc.

Dr. Valbona Bejleri, Supervisory Mathematical Statistician, USDA, NASS

Audience(s)

Department(s):

Mathematical Sciences