Predictive Analytics and ML Portfolio

The Hartford Financial Services

The problem

The Hartford Financial Services faced a pressing challenge within their Auto Insurance Data Science Personal Lines group. They lacked the essential expertise in R and Shiny required for developing modern predictive models.

Our solution

To address this issue, they brought Jasmine from Daly Analytics on board as an intern/contractor. Jasmine swiftly immersed herself in the intricacies of the auto insurance industry, gaining invaluable domain knowledge. Leveraging her expertise, she initiated several critical initiatives:

She implemented unconstrained loss modeling using advanced techniques like XGBOOST, enhancing the accuracy of loss predictions
She developed a Shiny model monitoring tool to track key model performance metrics
To streamline and optimize processes, she wrote technical documentation, custom functions and internal R packages tailored to the specific needs of Home, Auto, and Decision Analytics teams

This enabled The Hartford to increase confidence in predicting pure premiums and expanded the machine learning knowledge of the team.