Calculate WAIC

[drbenvincent]

In order to help facilitate model comparison, we should calculate WAIC statistics.

• [x] compute the log likelihood values from the models (see #196).
• [x] create a WAIC class to do the calculation, hold the stats. Create WAIC objects in the Model.postSamplingActivities method
• [x] allow comparison between multiple model fits. Compute a WAIC table.
• [x] allow plotting of WAIC info from multiple models
• [x] add example of comparing multiple models
• [x] become certain that this is working correctly

[drbenvincent]

What we have...

A model comparison table

Model comparison figure

This is at least doing vaguely the right thing! But I will check over to see if I'm calculating everything properly.

Things still to do:

• [x] Check dSE calculation
• [x] Calculate Akaike weights for a new weights column

[drbenvincent]

Now we have a comparison table:

and a nicely formatted WAIC comparison figure:

see demo_WAIC.m for en example