ResNCM: Conducting fairness analysis in STEM academic performance using Neural Causal Models

The data is available at Kaggle and in the cleaned, pre-processed data is available in the folder data.

To train a standard feed-forward NCM and compute counterfactual queries on the depression dataset, run all cells in depression.ipynb.

To train a standard ResNCM and compute counterfactual queries on the depression dataset, run all cells in depressionresnet.ipynb.

To train a standard feed-forward NCM and compute counterfactual queries on the exam scores dataset, run all cells in exam_scores.ipynb.

To train a standard ResNCM and compute counterfactual queries on the exam scores dataset, run all cells in exam_scores.ipynb.

The baselines and the data cleaning/pre-processing code can be found in the ipynb filed titled baselines.ipynb

Name		Name	Last commit message	Last commit date
Latest commit History 1 Commit
data		data
src		src
README.md		README.md
ResNCM.pdf		ResNCM.pdf
baselines.ipynb		baselines.ipynb
depression.ipynb		depression.ipynb
depressionresnet.ipynb		depressionresnet.ipynb
df_dep_test.csv		df_dep_test.csv
df_dep_train.csv		df_dep_train.csv
df_dep_val.csv		df_dep_val.csv
df_exam.csv		df_exam.csv
exam_baselines.ipynb		exam_baselines.ipynb
exam_scores.ipynb		exam_scores.ipynb
exam_scoresresnet.ipynb		exam_scoresresnet.ipynb
exp.ipynb		exp.ipynb
requirements.txt		requirements.txt

Provide feedback