rahmanidashti / calibrationfair

Calibration Fairness

CalibrationFair

Calibration Fairness

Dataset Preprocessing Pipeline

1. Downloading datasets raw file (datasets/DS_NAME/raw)
2. Collecting required features and mapping IDs to range 0 - N (the number of users or items). To do this we provide each dataset a specific notebook (datasets/DS_NAME/DS_Name_dataset.ipynb)
   • Output: Here we create two file, one for the rating data which show a user's rating on an item, and another one is cat file. The cat file indicate the category of each item (datasets/DS_NAME/DS_NAME_data_map.txt and datasets/DS_NAME/DS_NAME_cat_map.txt).
3. dataset.ipynb ---> datasets/DS_NAME/DS_NAME_data.txt and datasets/DS_NAME/DS_NAME_cat.txt
4. GoogleDrive/0_dataset_in_use.ipnb ---> Train, Test, Category, Inters
5. User Grouping (user-groups): 5%
6. Item Grouping (item-groups): 20%

Datasets

• ClothingFit: 5-core
• MovieLens1M: 10-core