Online Program

Classifying Risks for Underwriters: A Case Study in Risk Classification using Unsupervised Learning. (303927)

This presentation is a case study in how the challenge of correctly classifying a professional liability risk was addressed.

Underwriters help insurance companies select and maintain growing and profitable books of business by making decisions about acceptable risks. For medical professional liability, much of the decision-making process rests on a physician’s self-reported information. This is then corroborated as an underwriter pulls information from various sources, such as current and historical insurance policies, and evidence of practice (e.g. letterheads, website information, marketing materials). From the customer’s point-of-view, there is a desire for a more simplified process, with near instant results that provide a seamless process from quote to binding.

The fundamental underwriting challenge is to correctly classify the physician’s professional liability risk, have a near instant response quantifying the risk while providing a set of metrics that will assist in more accurate pricing for the risk. In response to these challenges, an underwriting prototype solution will be discussed that combines responsiveness and accuracy by integrating machine learning tools, software engineering, and end-use developers