[ASK] LibffmConverter - are the fit and transform function similar to Sklearn function such as OrdinalEncoder?
#1,818 opened on Sep 14, 2022
Repository metrics
- Stars
- (17,706 stars)
- PR merge metrics
- (Avg merge 6d 16h) (10 merged PRs in 30d)
Description
Description
I am splitting my dataset into Train, Validation, Test dataset.
For sklearn, when using OrdinalEncoder, the fit function will only be performed on Train dataset, I can use the same OrdinalEncoder fitted object to transform on unseen data(in this case Validation and Test dataset) without fitting again, this will preserve the same encoding and unseen data will be encoded as -1.
Does LibffmConverter perform the same way by simply fitting only the Train dataset and I can use the fitted object to transform other unseen(Validation and Test) dataset?
Example: converter = LibffmConverter().fit(train_df, col_rating='rating') train_df_new = converter.transform(train_df)
valid_df_new = converter.transform(valid_df). # preserving train fitted dictionary mapping and handle unseen data? test_df_new = converter.transform(test_df). # preserving train fitted dictionary mapping and handle unseen data?