A Feasibility Study on Automating the Automotive Insurance Claims Processing

In this paper, we present a feasibility study in automating the insurance claims processing system related to automotive sector. In automobile insurance claims processing, usually an adjuster is sent to specific location from where the claim has been registered to physically inspect the validity of the claim. Sending adjusters for each and every such claim imposes a cost factor. With the increased number of such claims it often becomes very difficult to physically inspect each claim with a limited number of adjusters. Here we categorize certain claims where physical inspection is unnecessary and these claims can be processed without sending any adjuster (we call these claims as ‘fasttrack’ claims). The challenge of labeling any claim as a ‘fast-track’ claim is that in the past historical data no claim has been identified as ‘fast-track’ claim. The domain experts have certain knowledge about ‘fasttrack’ness of a claim, however, it is very difficult to articulate that knowledge in the form of crisp rules. Rather if certain rules are articulated then the domain experts can validate whether such rules agree with their expertise or not. Therefore, our objective is to label a subset of claims as ‘fast-track’ claims and extract the rules for labeling any new claim as ‘fasttrack’ or not such that the rules must agree with the domain knowledge of the experts. We consider the delay in claims processing as an indicator variable, and learn decision tree structures for various different thresholds on the delay. We arrive the final decision tree structure where the total cost of false positive and false negative samples is minimized on the labeled data. Finally we validate the rules extracted by the decision tree with domain experts and observe that the decision tree brings out the domain expertise as the top set of rules. 15th International Conference on Management of Data COMAD 2009, Mysore, India, December 9–12, 2009 c ©Computer Society of India, 2009