The Graduate Student Seminar Speakers for Tuesday, May 9 will be Skyler Seto and Wenyu Zhang.
Skyler will present "Automatic Bayesian estimation of ARMA orders":
Abstract: An obstacle in employing ARMA models for time series modelling is choosing the correct autoregressive and moving average lag orders p and q. This paper proposes ABARMA, a highly practical, accurate, and automatic procedure for determining the values of p and q through Bayesian model selection. We determine model order by performing a change of variables and subsequently approximating the Bayesian evidence using the Laplace method. In simulations, ABARMA performs better or as good as competing methods in all cases. ABARMA exhibits consistent accuracy rates in different noise levels, and becomes increasingly accurate with more observations. In addition, we evaluate ABARMA on five real-world datasets.
Wenyu will present "Approximate Nonparametric Change Point Detection Procedure with Pruning":
Abstract: Change point analysis is a statistical tool to attain homogeneity within time series data. We propose a pruning approach for approximate nonparametric estimation of multiple change points. This general purpose change point detection procedure, cp3o, approximates the goodness-of-fit metric and applies a pruning step to the dynamic program to greatly reduce the search space and computational costs. A large class of existing goodness-of-fit change point objectives can immediately be utilized within the framework.