10.6084/m9.figshare.5691058.v1 Benoit Lamarsaude Benoit Lamarsaude Artificial intelligence applied in claims management: Bet on the right customer with claims satisfaction predictive modeling figshare 2017 Non Life Insurance Claims Management Artificial Intelligence Machine Learning Game Theory Markov Chains Entropy Insurance Studies Knowledge Representation and Machine Learning 2017-12-17 21:52:33 Thesis https://figshare.com/articles/thesis/Artificial_intelligence_applied_in_claims_management_Bet_on_the_right_customer_with_claims_satisfaction_predictive_modeling/5691058 <div>Insurance companies suffer from loss of customer consecutively to claims. Only a small portion of dissatisfied customer expresses themselves, creating difficulties in establishing a long term relationship. Increase customer loyalty is a major subject for insurers, because they have to maintain a minimum portfolio size and acquiring new</div><div>clients is more expensive than retains the existing. In this work we use artificial intelligence techniques to assess and manage the customer satisfaction when a claim occurs. Artificial intelligence offers a framework for prediction and decision making. A Markovian analysis based model which describes the claims management workflow is developed. The model provides predictions on customer satisfaction. Game theory is used to manage the predictions accordingly to insurer’s customer relationship strategy. The predictive model has been tested on real data from motor claims and on various machine learning algorithms. A game theory model describing relations between an insurance company and insured following a mismanaged claim is proposed and discussed. The results show that claims satisfaction redictive modeling could be used to develop customer loyalty through claims management. Directions for future works are proposed.</div>