free-soellingeraj
commented
3 years ago - copied prophet to a new folder called skaters/nproph
- refactored all the names proph > nproph (capitalizations, etc…)
- changed the filenames proph > nproph
- implemented test file tests/nproph/test_nproph.py copied from tests/proph/test_proph.py
- change skaters/nproph/nprophparams.py (adding defaults from output of ?NeuralProphet )
- change m.add_regressor(...) to m.add_lagged_regressor
- m.fit and m.create_future_dataframe take the same parameters but in a different pattern than with Prophet : both require df as input and m.fit takes freq (not create_future_dataframe)
- Commented out the stuff that adds the a_cols to future because it was already in future
- Commented out the stuff running the skater recursively for all exogenous y
microprediction