Split_by_idx returning out of bounds error?

Part 1 (2019)
1

I’m getting out of bound errors using split_by_idx after subsampling my df.

For example.

  • Split df into trn and val, get trn.index, val.index, passes to split_by_idx works fine.
  • After this step, I apply some criteria to reduce trn and val, recreate trn.index and val.index, and reconstruct a reduced df. I pass the reduced df and new trn.index and val.index, and I get an out of bounds error.