715 North Pleasant Street
Amherst, MA 01003
My research focuses on non-parametric and flexible parametric models for prediction and classification tasks with time series data that are relevant to public health. In the last few years, this work has centered on forecasting infectious diseases such as Dengue fever, influenza, and COVID-19. I have developed forecasting models using kernel conditional density estimation, copulas, and hierarchical splines or Gaussian processes, as well as ensemble methods to combine forecasts from multiple individual models. In my PhD dissertation I developed approaches to classifying physical activity using accelerometer data, and using conditional random field models in combination with random forests or gradient tree boosting.