Abstract
ChiRP: Chinese Restaurant Process Mixtures for Regression and Clustering
Highlights
These are common tasks in biomedical research
Chinese Restaurant Process (CRP) mixtures (Blackwell & MacQueen, 1973; Ferguson, 1973) are a class of Bayesian nonparametric models that can be used for robust regression modeling and clustering problems
Flexible machine learning methods exist for such problems, but they focus on predictive accuracy, making them inadequate for biomedical research applications where inference and interval estimation are of interest
Summary
ChiRP is a Monte Carlo Markov Chain (MCMC) implementation of Chinese Restaurant Process (CRP) mixtures in R. CRP mixtures (Blackwell & MacQueen, 1973; Ferguson, 1973) are a class of Bayesian nonparametric models that can be used for robust regression modeling and clustering problems. These are common tasks in biomedical research. Clustering often involves pre-specifying the number of clusters - typically unknown to the researcher. Predictions are formed by ensembling over the local cluster-specific regression models. This fully Bayesian procedure produces an entire posterior distribution for both the cluster assignments and predictions - allowing for both point and interval estimation
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have