Samir Khan Email & Phone Number

New York, Ny, Us

Stanford, Ca, Us

Brooklyn, Ny, Us

Proposed novel methods for decomposing long-term treatment effect estimates using Shapley values, allowing for the use of complex models to estimate customer lifetime value while preserving interpretability. Engineered features and explored model segmentation strategies for the lifetime value model, and doubled the model' R-squared.s

New York, Ny, Us

Developed a method to assign users a number of free articles for the month before hitting paywall. Trained random forests to predict user engagement and subscriptions, and implemented new causal estimators for backtesting. In backtesting, the resulting method generated approximately an additional 1000 subscriptions per month and $20,000 of additional revenue per month over baselines.

Worked with the Data for Good team to improve estimates of the number of people displaced by natural disasters. Added new features to a propensity model to improve estimates in sparsely populated areas and compared various approaches for building confidence intervals on both runtime and coverage.

Conducted research on high-dimensional statistics and statistical machine learning

Applied statistical methods to simulate elections and detect gerrymandering

Doctor Of Philosophy - Phd, Statistics

Bachelor'S Degree, Mathematics