Upper Division Course Projects at Harvard and CU Boulder

To provide a sense of what I’ve worked on other than my research in the National Studies on Air Pollution & Health Group and Earth Lab, here is a list of my most substantial projects that resulted in course papers but not peer-reviewed publications:

• Reviewed state-of-the-art offline reinforcement learning algorithms
• Investigated the usefulness of Bayesian Additive Regression Trees (BART) for estimating uncertainty in wildfire smoke prediction
• Explored methods to deal with measurement error (noise) in longitudinal (correlated) data
• Used distributed lag nonlinear models to investigate associations between air pollution and mortality
• Explored the evolution of discourse surrounding global public health surveillance and information systems, including future opportunities and risks in implementation
• Used network methods, spatial and socioeconomic data to model migration between U.S. cities
• Used data assimilation methods to model seasonal flu cases in Denver, CO
• Used spatial kriging to improve estimation of air quality from a random forest model
• Used satellite imagery to analyze trends in air quality over Southeast Asia
• Used a hierarchical model and change-point analysis to predict energy usage from temperature in U.S. cities. My model was the class’s most accurate on unseen data
• Used a convolutional neural network to predict sociodemographic information based on satellite images of US cities