Is there support for string feature values with preprocess_one_hot_encoding()? about mlpack HOT 5 CLOSED

Okay, I added the support in #3487. Once that is merged, and once a new version of mlpack is released, you should be able to pass pandas DataFrames in to the binding. Note that under the hood, what's actually happening is pretty much exactly the same as your workaround: we convert string values to integer values, then pass it to the one-hot encoding function.

from mlpack.

thank you @rcurtin !

from mlpack.

Sorry for the slow response here! Right now preprocess_one_hot_encoding() doesn't support categorical features... but it seems obvious that it should. I think it should be a simple code change (basically the input type of the binding just needs to be changed from a matrix to a categorical matrix), but I haven't had a chance to implement it yet. Anyway, it's on my list, and I'll post an update when I can make that change (or if someone else gets to it before I do feel free 👍).

from mlpack.

Sounds good, thanks for the clarification and update!

from mlpack.

resolved in #3487

from mlpack.