Request to use polars dataframe package directly

[jnhyeon]

Thank you for creating a great package.
polars is a next-generation high-performance dataframe package that will replace pandas.
After converting polars to pandas using to_pandas(), you can use the feature-engine package.
However, it would be even better if you could directly process polars dataframe in the feature-engine package!

Have a nice week :)

[solegalli]

Hey @jnhyeon

I agree with you: adding support for polars is a much needed improvement.

It's also a major undertaking because Feature-engine was designed with pandas in mind, so most of its functionality is based on pandas.

I'll need to get myself familiar with polars first, to get a clear view of the workload needed, or find someone with expertise that could take this forward.

So in principle, I agree, in practice, I am not sure how soon we can move forward with this.

Comparing the number of monthly downloads of pandas and polars, I think pandas is here to stay for the foreseeable future as well.

[hkrutzer]

In case you haven't seen it, https://github.com/narwhals-dev/narwhals may be helpful.

[jnhyeon]

Thank you so much for introducing such a great package.

Have a nice day :)

[gibou-g]

@solegalli got some ideas on this and keen to discuss further

[solegalli]

Hi @gibou-g

thanks for your message. Would be happy to discuss your ideas. I sort of started with the implementation here #965

Like @hkrutzer said, narwhals seems to be the path forward.