T. Qin

Results 17 comments of T. Qin

Sorry in advance to offer an somewhat pessimistic point of view. I am working on scientific computing in Rust so I really want to write down some of my reflections...

> Thank you for this comprehensive comment, @abstractqqq. > > As a maintainer, I welcome and value this critic. Would you like to give details about your experience with scikit-learn...

> I believe that this discussion has digressed, and I'd like to get things back on track. As a project, Linfa doesn't really follow grand plans. If anyone feels strongly...

I did an experiment using only Polars (not even Numpy) for my ML pipelines during 2022-2023. I rewrote almost all Transformers and metrics using Polars + some Rust. Let me...

@thomasjpfan You are exactly right it is about keeping the states. But I think it boils down to two steps: (1) learn from train. We can learn the mean, std,...

I totally agree that pipelines and transformers should be dataframe natives, with Numpy array being the second class.

A lot of the conversations have been about memory copy. Although I enjoy the technical discussions, I want to share some of my experience in credit modelling. Again, the focus...

I agree with everyone else and I am willing to put down a small bounty as well

> Ciao tq, thank you very much for the PR. Will try to review this ASAP. Not super expert on this branch of the codebase. In my second commit, I...

I explained in my commit's comment, here is short version: the "nan_case" tests are unnecessary in my opinion. This occurs when some feature is called on a null series (...