Here we studied two phase linear regression model with Bayesian approach. We also assume that at some point of time m, regression coefficient change from β1 to β2. For Bayes estimation of m, β1, β2 and 1/σ2 we used Gibbs sampling and MHRW (Metropolis Hasting Random Walk) algorithm. The effects of prior information on the Bayes estimates are also studied.
Two Phase Linear Regression Model, Change Point, Gibbs Sampling and MHRW Algorithm