It's not super easy to port a library with domain-specific ML/DS algorithms built around Pandas, such as accepting Pandas dataframes as arguments to its functions and class constructors.