Samuel Wilson

Windows support

1

Forgive me, I’m ignorant of compilation nuances on different operating systems. Is windows support on PyPi possible? I see that windows support was removed in a PR about a year...

How does lightgbm calculate the initial score for multiclass models?

4

For each tree in the multiclass model, I assumed that lightgbm would start boosting from `logit(% of total classes)`. It appears to work this way for a binary model. However...

See if tree compilation will speed up impute_new_data

1

https://github.com/siboehm/lleaves

If save_models == 1, procedure will be fastest and use least memory when predictions are compiled and deleted at the start/end of imputations for each dataset.

Especially the mean matching parameter gathering.

Often useful to specify that values can only be imputed using values from the same group.

It finally happened. This breaks all the tests. Need to find another dataset with a good combo of binary, multiclass and numeric columns.

The matrix operations for weights are simple enough. Need to test how much this will slow down the package.