Pandas or Dask to create features for XGBoost?

#1

I am using Pandas to create lots of features for training in XGBoost.

It appears that Pandas executes its functions as a single process using only a single CPU core. All other cores are unused. That results in feature generation taking hours. Would it be a good idea to replace Pandas with Dask or Modin for creating features?

#2

Yes. In addition, cuDF is also a good alternative, if you have access to a recent NVIDIA GPU.