Fraud claims are estimated to cost insurers a billion euros annually. During our case you will apply state of the art Machine Learning techniques to classify an insurer’s policyholders into segments and thereby prevent possible fraudulent claims. Classification of policyholders is not an easy task as it takes place in a complex multi-dimensional environment with highly non-linear relations. You will firsthand find out the differences between the classification power of a traditional multinomial Logit model and an Artificial Neural Network. This case shows you ‘what’s under the hood’ of these algorithms through step-by-step finetuning of the parameters. Finally, you will translate the technical work performed and convince the board of an insurance company to implement your algorithm of choice.
Please note that this event is open for BSc-3, BSc-4 and (pre-)master students only.