Big mistakes of data analytics when applied to risk and insurance
Data analytics in insurance offers unparalleled opportunities for new product design, risk selection, distribution, and modernizing the customer experience. These changes are also likely to pose potential problems like, for instance, security considerations of cloud computing, model risk from use of predictive models and insufficient technological skills of human resources. In this talk, Montserrat Guillen will present her views on some fundamental mistakes in the implementation of data analytic solutions in the insurance landscape, from data protection and data curation to the misuse of predictive modelling. She will also discuss the dangers of perpetuating the work in silos which, for decades, has characterized actuarial departments. She will advocate for a multidisciplinary approach. She will also show that standard analytics and statistical methods have a tendency to concentrate on the average, while risk is about the rare and extreme cases. Moreover, the majority of the existing methods ignore the dependencies between the data observations currently available in company databases. Overcoming these mistakes will result in a more dynamic, competitive, and innovative industry.
Registration from 15:30. Parkings are available near the NBB (e.g. Q-Park Pacheco)