https://nd.psychstat.org/ 2026-05-28T08:27:16+00:00 https://nd.psychstat.org/ https://nd.psychstat.org/_media/wiki/dokuwiki.svg text/html 2016-01-24T14:48:26+00:00 Anonymous (anonymous@undisclosed.example.com) https://nd.psychstat.org/codes/basic_mdp_r_code ################################################################ # Code to fit a basic MDP model # # Be warned that this is bad code. I've tried to keep the code # understandable, rather than efficient. ################################################################ ################################################################ # # Part 1. A description of the model # # The model is a conjugate, normal-normal model # Variances are treated as known throughout # # The model is the basic … text/html 2016-01-24T14:48:26+00:00 Anonymous (anonymous@undisclosed.example.com) https://nd.psychstat.org/codes/mdp_regression_r_code # Supplement to MDP code for regression modelling. # Code relies on the MASS library # The model to be fit with the code below is a # traditional linear regression model, with one # exception. The errors are presumed to follow # an "arbitrary" continuous distribution. The # prior over this distribution is an MDP model. # Note that the parameter sigma^2 is not the # usual residual variance. Also, one must be # careful in interpretation of "beta_0". The # value of the mean … text/html 2016-01-24T14:48:26+00:00 Anonymous (anonymous@undisclosed.example.com) https://nd.psychstat.org/codes/modified_basic_mdp_r_code ################################################################ # # Function 2. A function to generate theta_i for the cluster # structure # # Tasks: Compute probabilities of joining cluster, beginning # new cluster # Generate cluster membership # Update s and n_i # Update k and theta_star if needed # ################################################################ fn.gen.theta.i.2 <- function(i,clust,prior,mu,data) { n.i <- clust$n.i theta.… text/html 2016-01-24T14:48:26+00:00 Anonymous (anonymous@undisclosed.example.com) https://nd.psychstat.org/codes/more_modified_basic_r_code # Modification of MDP code to study mixing at a more basic level # First, a Gibbs sampler without the remixing step. If this sampler # is run with a large value of M (so that there are nearly as many # clusters as observations in the data set, mixing is nearly as good # as with the sampler with the remixing step. If the value of M # is relatively small, there are lots of "repeats", as clusters don't # often move. This behavior is not uniform across the cases in the # data set, but dep…