Αναλογιστικά μοντέλα τιμολόγησης και ανάλυση ασφαλιστικών δεδομένων με χρήση του πακέτου R
Actuarial pricing models and analysis of insurance data using the R packet
Πέτρου, Μαρία - Ελευθερία
KeywordsΓραμμικά μοντέλα ; Γραμμική παλινδρόμηση ; Ασφάλιση ; Γενικευμένα γραμμικά μοντέλα ; Διαχείριση κινδύνου
In many sciences we encounter the necessity of creating statistical models, namely the models which observe and describe a mathematical relationship between one or more variables. Accordingly, in insurance science we can predict damages, loss reserving, insurance rates etc., with the help of historical data. This thesis investigates the pricing process of an insurance product based on certain features of the customer, using generalized linear models, and the data analysis with which the models are being created. In the first chapter, we analyze the simple linear model and later on we expand our results to multiple. In the second chapter there is an introduction into the generalized linear model and a presentation of its structure, the way of assessment of the parameters and the diagnostic checks which are required to be performed on our results. Ultimately, in the third and final chapter, there are implementations of the generalized linear models in car insurance. More precisely, there is a presentation of techniques of calculating the pure premium, with which we would charge a customer, based on their personal features and we suggest a suitable model. For our calculations we use R, which is widely used for statistical data analysis, as well as for fitting the generalized linear models.