Western Reserve Group
Analyzing the Causes of Personal Auto Premium Changes
Team: Tianyi Cai '20, Tammy Dinh '20, Jordan Kirsch '20, Margaret Odero (Ashesi '19)
Advisors: Jennifer Bowen and Nathan Fox (Mathematics)
A local insurance company performed a rate revision, and some customer segments experienced unexpectedly large changes in their premiums. The AMRE team was given a set of data, which they then cleaned and analyzed. The team clustered the data into customer segments with similar characteristics and determined which segment was most affected by the rate revision. They also identified the factors that caused the most change in this revision, and analyzed the profitability of these customer segments. They used the techniques: k-means Clustering, Gower’s distance, Partioning Around Medoids (PAM), Boruta feature selection, and Random Forest. The team presented their findings to the company for further analysis and decision making.