I often run into situations where I hope pandas were 50-100x faster. Dask can help, but introduces quite a bit of additional complexity. I'm also looking forw

I often run into situations where I hope pandas were 50-100x faster. Dask can help, but introduces quite a bit of additional complexity. I'm also looking forward to stricter data models than what pandas currently uses, in particular proper null support for all dtypes and less random type conversion.