Applying generalized linear models to insurance data analysis

V magistrskem delu je predstavljena teorija posplošenih linearnih modelov (GLM) in uporaba le-teh v zavarovalništvu. Na realnih zavarovalniških podatkih izbrane zavarovalnice s pomočjo modelov GLM segmentiramo zavarovance glede na rizičnost in s tem določimo premijo za zavarovanje avtomobilske odgovornosti, ki je prilagojena posamezniku. S pomočjo logistične regresije analiziramo prekinitve življenjskih zavarovanj in zgradimo klasifikacijski model, ki razvrsti zavarovance v dve skupinina tiste, ki so oz. niso nagnjeni k prekinitvi.The master\u27s thesis presents the theory of generalized linear models (GLM) and their application in the insurance industry. Based on the real insurance data of the selected insurance company, we use the GLM models to segment insured persons according to risk, thus determining the premium for individual motor third party liability insurance. We use logistic regression to analyze the termination of life insurance policies and we build a classification model that classifies policyholders into two groupsto those who are and who are not prone to lapse