deepskies/DeepDiagnostics

New Plot: CDF parity plot

Opened this issue · 6 comments

See STAN and SBC checks paper

definition: True CDF vs. predicted CDF

There should also be a CDF difference plot, which is analogous to the residual/difference plot discussed in #29 (comment).

(TrueCDF - PredictedCDF)/PredictedCDF vs. TrueCDF

These are the plots that should come out of CDF parity.

@voetberg are these plots made anywhere?

I don't Think so - closest is the CDF ranks plot and I don't think that's the same. Do you have a link to the papers referenced in the issue definition?