Closed njtierney closed 6 years ago
JAGS uses coda.samples
and so returns an mcmc.list
. I've got an example I can commit and push to a new branch that simulates data for a linear model and shows off the column selection and makes a plot.
I have added a vignette for some simulated data, fitting and predicting from a linear model, showing density plots with the parameter values (without using ggmcmc
).
Perhaps we ought to change the vignette so that we sample from three sets of chains and look at colouring/grouping the graphs by chain.
Sorry I've been a bit absent here!
I really like the vignette, great job :)
Simulated data is great for what we need to show, so that works well.
However, I wonder if maybe we could use some example data from somewhere local (data is more fun if it's from Australia / Brisbane, imo) - we can then use that data in the package, and then fit a model to that, and store the model as data in the package.
pull request merged, I'll change the name on the trip home
It would be cool if we had a couple of datasets that were from different kinds of mcmc output - for example, I'm not sure if
mcmc.list
is the class for STAN and JAGS objects.