Statistical Inference for ML Fairness Metrics and Application to Recidivism Risk Forecasting

Description: Developing statistical inference procedures for ML fairness metrics defined in peer-reviewed study. Training various machine learning algorithms (Logistic Regression, LDA, Random Forest, Gradient Boosting) to forecast recidivism and evaluating model bias using developed inference procedures. Planning to submit to IEEE Big Data Conference.

GitHub Link

Utility Function for NFL Pass Plays and Defender Value Estimation

Description: Developed a utility function to quantify expected reward of passing plays. Trained ML algorithms to estimate pass completion probability and projected yardage gain after the catch. Analyzed trends in utility and marginal utility of defenders.

GitHub Link

Loan Repayment Prediction

Description: Conducted extensive exploratory data analysis of lending data from Lending Club. Used statistical learning methods and neural networks to predict the likelihood of loan default.

GitHub Link

Secure Computing Environment for Predictive Modeling and Generative AI

Description: Used cloud computing to create secure environment for predictive and generative modeling. Developed random forest classification algorithm to forecast whether a website visitor would become a potential customer.

GitHub Link

Drivers of Disposable Personal Income Growth

Description: Built a dataset of macroeconomic indicators and disposable personal income. Conducted regression analysis to predict disposable income growth using macroeconomic indicators. Addressed multicollinearity, heteroscedasticity, and normality violations.

GitHub Link