Density Regression

Density regression aims to model the entire conditional probability distribution of an outcome variable given predictor variables, going beyond simple point predictions. Current research emphasizes developing efficient and accurate methods for estimating these distributions, focusing on generative models like flow-based networks and Bayesian deep learning approaches to improve uncertainty quantification. These advancements are impacting diverse fields, enabling better uncertainty estimation in applications ranging from scientific modeling (e.g., microbiome data generation) to improved spatial analysis (e.g., crime hotspot prediction) and robust predictions under data shifts.