The main thrust of the proposed integrated research and education project will be the development of major enhancements to Bayesian Additive Regression Tree (BART) methodology for modern statistical problems. These enhancements will serve to address the growing need to address questions concerning complex processes underlying large modern multivariate data. Such questions arise, for example, in the study of biological, medical and physical systems, and in the data mining of massive datasets for business and financial analytics. In many applications, nonlinearity and high-dimensionality along with computational costs limit the scope for inferential investigations with standard Bayesian tools, while machine learning algorithms lack the capacity needed for a comprehensive statistical analysis. A flexible, fully Bayesian procedure resting only on minimal assumptions, BART has already proved itself to be broadly effective at discovering low dimensional signal hidden in high dimensional data while providing full posterior inference for full uncertainty quantification. The proposed enhancements to BART will serve to further unleash the potential of BART in important new directions for contemporary statistical analysis.
|Effective start/end date||7/15/19 → 6/30/22|
- National Science Foundation (NSF): $159,997.00
Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.