Final edits BART

[catavallejos]

• For prediction, only show CIF in order to be consistent with others.
• Explore convergence before showing predictions
• What is the purpose for the following lines? Predictions were shown already. Is this to show how predictions could be made when the test dataset was not in the original crisk2.bart() call? If so, explain.

https://github.com/KarlaMonterrubioG/Competing_risks/blob/88bc2a15f0429b01244d2232af591c2a75522a83/Source/Discrete_specification.Rmd#L225-L260

I added a sentence, was to clarify that inn reality you always need to run the model with test=train as you need that to revise the MCMC chains. However, you can do predictions afterwards. If you directly use test=test you get info about the chains at test (which is odd).

• Edit output in second formulation to be consistent with what was shown for first one

[KarlaMonterrubioG]

I pushed some changes on BART, but note that the html file is not updated - takes a bit of time and for some reason my lap keeps crashing due to memory because of all the chains. Will update as soon as I am able to run