dirty-cat: machine learning on dirty data

[jovan-stojanovic]

Hi,

You might want to add the dirty_cat package to your list. It has some features that seems to fit to your interests:

• fuzzy_join: fuzzy joining tables on numerical, string or mixed columns.
• FeatureAugmenter: fuzzy joining multiple tables at the same time.
• deduplicate: deduplicating dirty categorical variables.

And it's scikit-learn compatible.

Cheers.

[J535D165]

Looks very interesting. Would you consider a PR?

[jovan-stojanovic]

Sure, I can work on it.