Sparse Event Modeling with Hierarchical Bayesian Kernel Methods
Abstract
The research objective of this proposal was to develop a predictive Bayesian kernel approach to model count data based on several predictive variables. Such an approach, which we refer to as the Poisson Bayesian kernel model, is able to model the rate of occurrence of events (and subsequently, their likelihood of occurrence) based on historical evidence of the counts of previous event occurrences. The novel Bayesian kernel methods made use of: (i) the Bayesian property of improving predictive accuracy as data are dynamically obtained, and (ii) the kernel function which adds specificity to the model and can make nonlinear data more manageable. Early results show that the Poisson Bayesian kernel model is more effective than the Poisson generalized linear model at modeling rates of occurrence especially for small datasets where regression-based methods often fail. The ability to model sparse data sets represents a positive step in modeling low-likelihood events often encountered in risk analysis.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 05, 2016
- Accession Number
- AD1008781
Entities
People
- Kash Barker
Organizations
- University of Oklahoma