Software



MORF The MOOC Replication Framework (MORF) is an open-source platform for reproducible machine learning research on privacy-protected data. MORF uses containerization to perform secure, scalable computation on MOOC datasets representing millions of students, providing unprecedented access to a multi-instutitional dataset for understanding learning at scale.
ABROCA The R package for computing Absolute Between-ROC Area (ABROCA). ABROCA is a metric proposed in our LAK19 paper for measuring predictive model bias across demographic groups, and is particularly useful when predictive models will be used for many different downstream tasks (and therefore fairness across all possible prediction thresholds is desired).
auctestr The auctestr R package enables statistical testing of the Area Under the Receiver Operating Characteristic Curve (AUC ROC), a model performance metric which is commonly used but is rarely subjected to statistical evaluation, despite longstanding and well-known statistical properties of the AUC statistic.


I've written lots of other code, and occasionally contribute to other open-source projects. You can find some of this work on my GitHub profile.