Claims Severity Modeling
This study is presented as a portfolio of three projects, two of which were a part of my summer internship at CNA Insurance, Chicago and one was a part of the course STA 663: Statistical Computation.
Project 1, Text Mining Software, aimed at building an efficient text mining software for CNA Insurance, in the form of an R package, to mine sizable amounts of unstructured claim notes data to prepare structured input for claims severity modeling. This software decreased run-time 30 fold compared to the software used previously at CNA.
Project 2, Workers’ Compensation Panel Data Analysis, aimed at tracking workers’ compensation claims over time and pointing out variables (particularly medical) that made a claim successful in the long run. It involved creating a parsimonious Mixed Effects model on a panel dataset of Workers’ Compensation claims at CNA Insurance.
Project 3, Infinite Latent Feature Models and the Indian Buffet Process (IBP), used IBP as a prior in models for unsupervised learning by deriving and testing an infinite Gaussian binary latent feature model. An image dataset was simulated (similar to Griffiths and Ghahramani (2005)) to test the model.
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 United States License.
Rights for Collection: Masters Theses