It would be great to provide a common interface in Python and R using `pandas.DataFrame` and `data.frame`, respectively. Options: - Wrap `pytreecat` in `reticulate` (as done by `rstudio/tensorflow`). Use either `feather` or upcoming dataframe conversion https://github.com/rstudio/reticulate/issues/2 - Reimplement the core of treecat in C++ and wrap in Rcpp