predictions dmatrix and remove padding

vincentarelbundock / pymarginaleffects

GNU General Public License v3.0

47 stars 8 forks source link

predictions dmatrix and remove padding #97

Closed LamAdr closed 2 months ago

LamAdr commented 2 months ago

This fixes #83 and #95 with @RoelVerbelen's suggestions.

all padding code is boldly removed. Given that I barely understands what it does that might be naive.
we get exog from dmatrix using design_info, not formula.
the NA_action="raise" is used to flag missing values. I found Patsy's error message pretty informative so I did not catch it.

Please tell me if I should change anything

vincentarelbundock commented 2 months ago

I think the errors are because of differences in plots. If you have time, could you see what's up with that? Are the plots really different, or are they minor pixel differences?

(that should probably be a separate PR that we merge in here)

LamAdr commented 2 months ago

Yeah looks like minor pixel differences. Should i regenerate the images?

vincentarelbundock commented 2 months ago

Yep, if there's really no substantive difference, then feel free to re-generate. Thanks!

vincentarelbundock commented 2 months ago

Thanks!