conditional_after in WeibullAFTFitter.predict_expectation()

[sve007]

Hi all,

Can someone enlighten me why e.g. the WeibullAFTFitter().predict_expectation() function does not have the parameter conditional_after? When using predict_median for these models I do have this option. Is there a good mathematical explanation or is it something that is not (yet) implemented?

Thanks

[CamDavidsonPilon]

It's not implemented, but the gist of it is that it would rely on an approximation algorithm. To get around this, try:

from lifelines.fitters import ParametricRegressionFitter
super(ParametricRegressionFitter, waft).predict_expecation(..., conditional_after=...)

Hacky, but I can make this better in the next release.

[sve007]

Thanks, I will give this a try. Good to know that there is not a mathematical reason that this wouldn't be possible.