CFE 2019: Start Registration
View Submission - CMStatistics
Title: Bayesian CUSP catastrophe model for sudden changes Authors:  Zhuoqiong He - University of Missouri (United States) [presenting]
Abstract: The cusp catastrophe model uses a discontinuous nonlinear function to predict sudden changes. Due to the complexity of the discontinuous nonlinear relationship, there are some issues in fitting the statistical cusp regression model, i.e., gradient based optimization methods no longer work. We have developed a Bayesian method for the cusp regression model and used the posterior mean to obtain estimates of the parameters. The partial swarm optimization algorithm is used to speed up the convergence of the Markov chain Monte Carlo algorithm. The simulation study shows that the Bayesian method yields a better estimate than both the maximal likelihood estimation and the traditional stochastic differential equations method under the Maxwell convention.