pandas>=0.23
scikit-learn>=0.23
sparse>=0.9
dill
xlcompose
patsy
