Skip to content

Commit

Permalink
make data leakage bug to load_data_from_folder
Browse files Browse the repository at this point in the history
  • Loading branch information
codeKgu committed Nov 16, 2020
1 parent f6aad22 commit 8b3d0ee
Showing 1 changed file with 7 additions and 3 deletions.
10 changes: 7 additions & 3 deletions multimodal_transformers/data/load_data.py
Original file line number Diff line number Diff line change
Expand Up @@ -108,10 +108,14 @@ def load_data_from_folder(folder_path,
data_df = pd.concat([data_df, cat_df], axis=1)
categorical_cols = cat_feat_processor.feat_names

train_df = data_df.loc[train_df.index]
len_train = len(train_df)
len_val = len(val_df) if val_df is not None else 0

train_df = data_df.iloc[:len_train]
if val_df is not None:
val_df = data_df.loc[val_df.index]
test_df = data_df.loc[test_df.index]
val_df = data_df.iloc[len_train: len_train + len_val]
len_train = len_train + len_val
test_df = data_df.iloc[len_train:]

categorical_encode_type = None

Expand Down

0 comments on commit 8b3d0ee

Please sign in to comment.