I’m running a polynomial regression between a multimodal temperature variable “temperature” and a normally-distributed variable “gdp”. I am having trouble understanding why the sampling chains in my regression analysis converge to two distinct values for the regression coefficients. Model code below…

My strong suspicion is that each solution the posterior finds is associated with a mixture component of temperature. You can dig into this by looking at az.plot_pair, coloring the scatter plots by chain. You will also notice that the regression solution corresponds to the “peakier” of the two compo…

Chains converge to different results

Questions version agnostic

jessegrabowski November 6, 2023, 1:55pm 4

Why are you interested in explicitly modeling an exogenous input to your regression?

Topic		Replies	Views
Gaussian Mixture of regression Questions	5	4054	November 10, 2017
Mixture regression Questions	2	1010	January 29, 2019
Mixed multivariate Gauss distribution Questions	39	4948	April 12, 2018
Mixture of hierarchical model Questions	8	3079	May 31, 2018
Linear regression model: Three chains work, one chain never moves Questions	16	1660	April 30, 2019

Chains converge to different results

Related topics