Titolo della tesi: Pricing sophistication in travel insurance through GLM, Machine Learning and text mining data enrichment in a Covid-19 world.
The aim of this dissertation is to propose new pricing techniques for the travel insurance products, in light of a new world situation brought by Covid-19 and a travel marked disrupted by it.\\
Travel insurance is a niche product characterized by high margins and under developed pricing techniques, that differs from other non-life insurance product in term of duration, price, mode and timing of purchase etc.
Before applying the various statistical models tested to estimate the frequency of travel cancellations, ranging from classical GLM to sophisticated machine learning techniques, a text mining approach to data enrichment is presented in consideration of the amount of textual information that this type of business is able to produce.
The present work shows that, thanks to the newly produced variable and the applied models, we are able to produce a better segmentation for the travel cancellation risk compared to the pricing methodologies currently used by the biggest players of the market for these products.
Finally, a new product covering financial indemnity and assistance to Covid-19 patients is presented, whose frequency is dynamically estimated with a SIR model as no historical data could yet be used.