Praveen Nair‍

(Nair rhymes with 🔥)


Patterns of Fairness in Machine Learning

correlation heatmap of raw metric values

This was my data science senior capstone project, in which two other students and I, under the supervision of Professor David Danks, looked at how different classifiers performed on multiple canonical fairness datasets, for a series of different metrics of algorithmic fairness. The project was designed to be user-extensible to new datasets and metrics. The GitHub repo for our project is here, and the report for the project is available here.

Logistic Regression Penalizing Demographic Disparities

correlation heatmap of raw metric values

Final project for CSE 203B, Convex Optimization, at UCSD. Extended work from Bechavod and Ligett (2017) to develop additional fairness penalizers in optimization formulation of logistic regression. Derived dual formulation and solved with CVXPY. Also, the first time I ever typeset a bunch of math in LaTeX, which isn’t an achievement, but it sure felt like it. Report available here.