Pricing models for car insurance traditionally use variables related to the policyholder and the insured vehicle (e.g. car brand and driver age) to determine the premium. This can lead to situations where policyholders belonging to a group that is seen as carrying a higher risk for accidents wrongfully get a higher premium, even if the higher risk might not necessarily apply on a per- individual basis. Telematics data offers an opportunity to look at driving behavior during individual trips, enabling a pricing model that can be customized to each policyholder. While these additional variables can be used in a generalized linear model (GLM) similar to the traditional pricing models, machine learning methods can possibly unravel non-linear connections between the variables. Using telematics data, we build a gradient boosting model (GBM) and a neural network (NN) to predict the claim frequency of policyholders on a monthly basis. We find that both GBMs and NNs offer predictive power that can be generalized to data that has not been used in the training of the models. The results of the study also show that telematics data play a considerable role in the model predictions, and that the frequency and distance of trips are important factors in determining the risk using these models. / Prissättningsmodeller för bilförsäkringar använder traditionellt variabler relaterade till försäkringstagaren och det försäkrade fordonet (t.ex. bilmärke och förarålder) för att bestämma försäkringspremien. Detta kan leda till situationer där försäkringstagare som tillhör en grupp som anses bära på en högre risk för olyckor får en felaktigt hög premie, även om den högre risken inte nödvändigtvis gäller på en individbasis. Telematikdata erbjuder en möjlighet att titta på körbeteende under individuella resor, vilket möjliggör en prissättningsmodell som kan anpassas till varje enskild försäkringstagare. Ä ven om dessa variabler kan användas i en linjär modell liknande de traditionella prissättningsmodellerna kan användandet av maskininlärningsmetoder möjligen avslöja icke-linjära samband mellan variablerna. Med hjälp av telematikdata bygger vi en modell baserad på gradient boosting (GBM) och ett neuralt nätverk (NN) för att förutsäga frekvensen av olyckor för försäkringstagare på månadsbasis. Vi kommer fram till att båda modeller har en prediktiv förmåga som går att generalisera till data som inte har använts vid träningen av modellerna. Resultaten av studien visar även att telematikdata spelar en betydande roll i modellernas prediktioner, samt att frekvensen och sträckan av resor är viktiga faktorer vid bedömningen av risken med hjälp av dessa modeller.

Mémoire numérisé par la Division de la gestion de documents et des archives de l'Université de Montréal.

Mémoire numérisé par la Division de la gestion de documents et des archives de l'Université de Montréal

This thesis characterizes the products mandatory liability insurance of car traffic, car insurance and some additional insurances of cars. The main objective is the analysis of calculation and of criterias, that determinate the price of insurance. The end point is approach comparison to setting up the price by insurance companies on the Czech insurance market.

The main objective of this diploma thesis is the analysis of the products of motor insurance offered on the insurance market in the Czech Republic. The first part focuses on character and meaning of this products and clarification of some concepts. The next part describes current situation and development in the insurance market followed by the very topical stiff competition. The last part is devoted to the analysis of supply in the market of car insurance itself and its evaluation from the point of view of various parameters and criteria.

[pt] A estimação do risco em seguros de automóveis representa um difícil problema de regressão. As dificuldades vão desde a utilização de um grande número de variáveis discretas como explicativas, até a distribuição particular dos ruídos e uma quantidade expressiva de categorias com valores nulos e valores discrepantes. Supondo que os problemas de estimação estejam relacionados com a classificação do risco adotada pelo mercado, este trabalho propõe um método de classificação alternativo. O método desenvolvido foi baseado na técnica de análise fatorial, e no algoritmo de agrupamento de dados denominado fuzzy clustering system. Para avaliar a eficiência do método em solucionar os problemas de estimação, optou-se por utilizar o erro resultante da aplicação de modelos lineares generalizados. Ao final, o erro de estimação obtido diante da classificação proposta, foi comparado ao obtido diante da classificação usual de mercado. / [en] The estimation of car insurance risk rate represents a difficult regression problem. One of the difficulties of this problem is the use of a number of discrete independent variables and a specific error distribution that presents an expressive number of null and outlier values. Assuming that these estimation problems are related to the risk classification adopted by the insurance companies, this work proposes an alternative classification method. This method is based on factorial analysis techniques and on the algorithm known as Fuzzy Clustering System. To evaluate the efficiency of this method in solving the problems identified, the risk was estimated using generalized linear models. The errors from each model were obtained and compared between classifications.

The purpose of my Diploma Thesis is to lay out the philosophy of the internet and searching in fulltext locators and apply it to analysed environment of the insurance which is presented on the internet. Search Engine Optimization is nowadays the one faced with the activities of online marketing. Many of the views, opinions and qualitative contrasts meet and reach still bigger or smaller achievements in relation to developing locators. Diploma Thesis suggests at the fact, that conditions according to the locator decides, are changing, but the philosophy is still left. In this case and reason, the work proves efficiency of the activities by various case studies, which follow the philosophy and do not look just at the technical side of SEO.

This thesis deals with a specific area of telematics systems - intelligent transport systems for private vehicles. The main objective is to compare telematics services in the insurance industry in the Czech Republic and in selected foreign countries (UK, USA, Germany and Russia) and propose an extension of these services, which could be implemented in a particular society dedicated to the insurance industry. Proposal to extend the product service from the insurance telematics is drawn through two project plans. The main goal is accomplished through several other sub-objectives, in identifying the theoretical concepts related to these issues and overall definition of telematics for analysis of telematic services. Another sub-goal is to monitor the Czech market providers of navigation systems and vehicle monitoring, comparison and evaluation of the products aimed at this issue. Contribution of the work is mainly mapping market, what products and services are offered. The solution also helps to compare the chosen methodology through the evaluation criteria. Evaluation of this situation will be reflected as a recommendation for a specific provider from the insurance industry. Analysis of products and services is also beneficial for the customer, provides an overview of the products and their evaluation, which basically points out what is appropriate to focus on making these products.

L’aléa moral est une source de distorsion économique. La prédiction classique dans un cadre simple est qu’une meilleure couverture conduit à un effort moindre. Cette thèse étudie dans quelle mesure cette prédiction est ou non vérifiée empiriquement dans des cadres plus complexes. Le premier chapitre s’intéresse aux absences pour raison de santé. La politique étudiée est le jour de carence pour arrêt maladie dans la fonction publique de l’Etat en France. Cette politique de remboursement moins généreuse a notamment pour but de réduire l’absentéisme. Elle conduit à une baisse de la prévalence des absences de courte durée. Mais elle conduit aussi à une hausse de la prévalence des absences de longue durée. En conséquence, la prévalence de l’ensemble des absences pour raison de santé reste inchangée. Les deux chapitres suivants s’intéressent aux systèmes de bonus-malus d’un assureur automobile irlandais. Le deuxième chapitre s’intéresse à l’introduction d’un état très protecteur : la protection à vie du bonus. Cette protection est octroyée automatiquement et gratuitement aux assurés sous des conditions restrictives d’historique de sinistre et d’ancienneté. Comparé à la situation dans laquelle cet état protecteur n’existe pas, le taux de sinistre des assurés protégés augmente, tandis que le taux de sinistre des agents non protégés diminue, dans l’espoir d’être récompensés par la protection. L’existence de la protection est à l’origine d’un transfert intertemporel. Les assurés renoncent à de l’utilité présente en exerçant un effort supérieur, afin d’être récompensés par la protection et de profiter d’une utilité future plus élevée due à un effort moindre. Le troisième chapitre étudie la réaction juste après que l’assuré est récompensé par la protection à vie du bonus. Le taux de sinistre augmente immédiatement, mais seulement quand la protection existe depuis quelque temps. Cela suggère que l'effet d'un changement incitatif dépend de sa nature, mais aussi de son contexte. / Moral hazard is a source of economic distortion. The classical prediction in a simple framework is that a better coverage leads to a lower effort. This thesis studies the extent to which this prediction is empirically verified in more complex settings. The first chapter focuses on health-related absences. The policy under study is the one-day waiting period for sick leave in the French central civil service. This less generous reimbursement policy notably aims at reducing absenteeism. It leads to a decrease in the prevalence of short-term absences. But it also leads to an increase in the prevalence of long-term absences. As a result, the prevalence of all health-related absences stay unchanged. The two following chapters focus on bonus-malus systems used by an Irish car insurer. The second chapter focuses of the introduction on a highly protecting state: the lifetime bonus protection. This protection is granted automatically and freely to insurees under restrictive conditions on past claims and seniority. Compared to the situation in which this protecting state does not exist, the claims rate of protected insurees increases, but the claims rate of unprotected insurees decreases, in the hope of being rewarded with the protection. The existence of the protection induces an intertemporal transfer. Insurees waive present utility by exerting more effort, so as to be rewarded with the protection and to enjoy more future utility due to lower future effort. The third chapter studies the reaction just after the insuree is rewarded with the lifetime bonus protection. The claims rate increases immediately, but only when the protection exists for some time. This suggests that the effect of an incentive change depends on its nature, but also on its context.

La recherche menée dans cette thèse propose une contribution à l’analyse du risque sur le marché de l’assurance automobile en France. Trois nouveaux axes sont présentés : le premier axe s’inscrit dans un cadre théorique de marché d’assurance automobile. Un modèle original de double asymétrie d’information est présenté. Le principal résultat qui en découle est l’existence de deux sortes de contrats d’équilibre : un contrat séparateur et un contrat mélangeant. Le deuxième point est lié à la prise en compte de la sinistralité passée dans l’étude de la relation risque - couverture. Des modèles bivariés et trivariés sont appliqués pour cette fin. Il en ressort que l’hypothèse de l’asymétrie d’information est vérifiée. Enfin, la troisième question soulevée dans cette thèse concerne l’application de la surprime aux jeunes conducteurs. Nous montrons par des modélisations économétriques de la sinistralité que la légitimité des assureurs à proposer quasi systématiquement des tarifs plus élevés aux jeunes conducteurs par rapport aux conducteurs expérimentés n'est pas toujours vérifiée. / This dissertation provides a contribution to the risk analysis on the French automobile insurance market. The objective of this thesis is threefold. The first aim relates to a theoretical framework applied on insurance market. An original model of double asymmetry of information is presented The main result that emerges is the existence of two kinds of contracts at equilibrium :a separating contract and a pooling contract. The second point concerns the past claims and the risk-coverage correlation. Bivariate and trivariate models are applied for this purpose. It results that the assumption of asymmetry of information is not rejected. The third issue is related to the over-premium that insurers apply quasi-systematically to the young drivers. We show, using econometric modeling, that this over-pricing compared to the experienced drivers’premium isnot necessary and its removal does not compromise the sustainability of the insurance company.