This is the repository for the paper Methodological Considerations for Predicting At-risk Students at the ACE22 conference.
In this work we investigate the potential bias when predicting students dropping out that arises from including already dropped out students in the training data.
For privacy reasons, unfortanetely we are unable to disclose the data used in this study. But the code that was used to obtain our results is available here in this repository.
The data_analysis.ipynb
and plots.ipynb
notebooks include the dropout data statistics and plots present in the paper
as well as the code that generated them.