Fair Cognitive Diagnosis Without Sensitive Attributes (FairWISA)

The code for the paper "Fair Cognitive Diagnosis Without Sensitive Attributes" (FairWISA).

The full version of the code is coming soon...

Data

PISA data is available at: PISA Data

SLP data is available at: SLP Data

Environment Requirements

• EduCDM
• torch: 1.13.1
• pandas: 1.0.1
• scipy: 1.4.1
• numpy: 1.21.6
• tensorboardX: 2.6
• scikit-learn: 0.22.1
• tqdm: 4.42.1

How It Runs

1. Train a Cognitive Diagnosis (CD) model (e.g., IRT) to get the model parameters file: model.params.
2. Execute the following command to generate the grouping matrix file: groups.params:

   python gengroups.py

3. Execute the following command to obtain the fair CD model:

   python test