Plots all trace plots for the phi parameters in all populations. For convenience, the \(F_{k}\) statistic is presented in place of the phi parameter, as this is the statistic users care about. \(F_{k}\) is defined as \(\frac{1}{1+phi_{k}}\).
plot_all_phi_trace(phi_mat, percent.burnin = 0, thinning = 1, population.names = NULL)
The k by ngen matrix of phi values estimated for all k
populations/individuals included in the analysis in each of ngen MCMC
generations.
The percent of the sampled MCMC generations to be discarded as "burn-in." If the
MCMC is run for 1,000,000 generations, and sampled every 1,000 generations, there
will be 1,000 sampled generations. A percent.burnin of 20 will
discard the first 200 sampled parameter values from that sample.
The multiple by which the sampled MCMC generations are thinned. A thinning of
5 will sample every 5th MCMC generation.
A vector of length k, where k is the number of populations/individuals
(i.e. k = nrow(counts)), giving the name or identifier of each
population/individual included in the analysis. These will be used to title the
k trace plots of the phi parameters estimated for each population/individual
in the beta-binomial model. If population.names is not provided
(i.e. population.names =NULL), a population index number will be used to
title the plot.
Gideon Bradburd
A trace plot is a basic visual tool for assessing MCMC mixing. If the chain is mixing well, the trace plot will resemble a "fuzzy caterpillar." If the trace plot has not plateaued, it is an indication that the chain has not converged on the stationary posterior distribution, and must be run longer. If the trace plot of a parameter exhibits high autocorrelation, the user may wish to either increase or decrease the scale of the tuning parameter on that parameter, to decrease or increase acceptance rates, respectively. If the chain appears to be bouncing between areas of "fuzzy caterpillar-dom," it may be an indication of a multi-modal likelihood surface.