search

microsoft / lightgbm-transform

Transformation library for LightGBM
MIT License
33 stars 8 forks source link

[feature request] Encodings for categorical features #21

Open guolinke opened 2 years ago

guolinke commented 2 years ago

Categorical encodings, like one-hot, frequency, and target encoding, are widely used in tabular data. Does this repo support the categorical features?

chjinche commented 2 years ago

Hi @guolinke thanks for asking! Currently, the repo just supports FreeForm2Parser. FreeForm2 evaluates a single, nested expression that returns a floating-point number, so categorical encodings have not been supported yet. However, as the repo supports parser customization, I think categorical transform is doable although may take some efforts. Could you please share some motivation and real needs behind the ask? Just want to get deeper understanding of our user needs.