This diploma project is focused on the estimation of incurred value of claim and probability of the claim remaining opened (not settled) in the specific stage of the insurance settlement process. The change of incurred value of claim means the change of settlement process stage. Generalized linear model is used for modelling these changes. Classical linear regression model also belongs into this theory, which is its special case, just with stricter premises. Generalized linear model among others allows solving the problem of heteroscedasticity in the unusual way using joint model. This model is applied in the practical part of this piece of work. Logistic regression is the part of the generalized linear model theory, which helps to model the probability of the claim remaining opened in this piece of work. The model outcome is presented in graphic way, especially the graphs containing probability that levels of given claim will occur in certain range.

The news media is an important institution for all democracies. It helps the citizens to keep informed and be able to take part of the public debate, but in recent years the gap between the active and the inactive news consumer has increased. Does it make any difference? In order to contribute to the field, this research paper is to make a quantitative analysis to look at whether people with a high consumption of news from the Swedish mass media differ in their attitude towards the EU. As an ordered logistic regression was not applicable when analyzing the categorical dependent variable, that are measuring attitudes towards the EU, three binary logistic regressions was instead used. The results show that individuals with a high consumption of news from the Swedish mass media have higher odds of having an opinion of a positive attitude toward the EU. The data used are however self provided and voluntary survey data,which contain various biases. The fact that it is only observed and not experimental data makes it impossible to estimate a causal effect, which instead is up to future research.

Credit scoring is important and rapidly developing discipline. The aim of this thesis is to describe basic methods used for building and interpretation of the credit scoring models with an example of application of these methods for designing such models using statistical software R. This thesis is organized into five chapters. In chapter one, the term of credit scoring is explained with main examples of its application and motivation for studying this topic. In the next chapters, three in financial practice most often used methods for building credit scoring models are introduced. In chapter two, the most developed one, logistic regression is discussed. The main emphasis is put on the logistic regression model, which is characterized from a mathematical point of view and also various ways to assess the quality of the model are presented. The other two methods presented in this thesis are decision trees and Random forests, these methods are covered by chapters three and four. An important part of this thesis is a detailed application of the described models to a specific data set Default using the R program. The final fifth chapter is a practical demonstration of building credit scoring models, their diagnostics and subsequent evaluation of their applicability in practice using R. The appendices include used R code and also functions developed for testing of the final model and code used through the thesis. The key aspect of the work is to provide enough theoretical knowledge and practical skills for a reader to fully understand the mentioned models and to be able to apply them in practice.

In this paper the hypotheses related to individual variables used for segmentation of Motor Third Party Liability (MTPL) insurance by Czech insurance companies are tested. Summary of papers focused on this topic and development of insurance market segmentation in European Union are presented in the first part of this thesis. The first part of this paper is extended by the analysis of actual MTPL segmentation in Czech Republic. The estimation of marginal effects of exogenous variables on probability of occurrence a claim is described in empirical part. For the estimation of parameters I use the logistic regression. Specific models for small and large claims are created. The most significant variables positively correlated with probability of occurrence a claim are engine capacity, young age and region Prague. The most significant variables negatively correlated with probability of occurrence a claim are historical car, old age, number of months without any claim and region South Moravia. My results are compared to the results of other papers.

The aim of this thesis is to analyse the material deprivation of the households with regard to the selected indicators, i.e. the costs that the household spends on housing, a region where the household is located, the number of the members and the dependent children in the household, age and sex of a head of the household, and economic activity and education level of the members of the household. The thesis aims not only to prove the dependence among the selected indicators but also to quantify this dependence by using the odds ratio. The individual effect of all variables was proven except of the one related to the number of the dependent children. It was also demonstrated that the factors constituting a threat for the households by a material deprivation rate vary by the different age groups. However, it can be concluded that across all the age groups, the material deprivation rate is determined by the sex of a head of the household, education level of the members of the household, and the costs that the household spends on housing.

The majority of documented phishing attacks have been carried by email, yet few studies have measured the impact of email headers on the predictive accuracy of machine learning techniques in detecting email phishing attacks. Research has shown that the inclusion of a limited subset of email headers as features in training machine learning algorithms to detect phishing attack did increase the predictive accuracy of these learning algorithms. The same research also recommended further investigation of the impact of including an expanded set of email headers on the predictive accuracy of machine learning algorithms. In addition, research has shown that the cost of misclassifying legitimate emails as phishing attacks--false positives--was far higher than that of misclassifying phishing emails as legitimate--false negatives, while the opposite was true in the case of fraud detection. Consequently, they recommended that cost sensitive measures be taken in order to further improve the weighted predictive accuracy of machine learning algorithms. Motivated by the potentially high impact of the inclusion of email headers on the predictive accuracy of machines learning algorithms and the significance of enabling cost-sensitive measures as part of the learning process, the goal of this research was to quantify the impact of including an extended set of email headers and to investigate the impact of imposing penalty as part of the learning process on the number of false positives. It was believed that if email headers were included and cost-sensitive measures were taken as part of the learning process, than the overall weighted, predictive accuracy of the machine learning algorithm would be improved. The results showed that adding email headers as features did improve the overall predictive accuracy of machine learning algorithms and that cost-sensitive measure taken as part of the learning process did result in lower false positives.

The purpose of this study was to investigate predicting student success on one example of a "high stakes" test, the Texas Assessment of Academic Skills Exit Level Exam. Prediction algorithms for the mathematics, reading, and writing portions of the test were formulated using SPSS® statistical software. Student data available on all 440 students were input to logistic regression to build the algorithms. Approximately 80% of the students' results were predicted correctly by each algorithm. The data that were most predictive were the course related to the subject area of the test the student was taking, and the semester exam grade and semester average in the course related to the test. The standards of success or passing were making a 70% or higher on the mathematics, 88% or higher on the reading, and 76% or higher on the writing portion of the exam. The higher passing standards maintained a pass/fail dichotomy and simulate the standard on the new Texas Assessment of Knowledge and Skills Exit Level Exam. The use of the algorithms can assist school staff in identifying individual students, not just groups of students, who could benefit from some type of academic intervention.

Orientador: Elizabeth Nogueira Gomes da Silva Mercuri / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Educação / Made available in DSpace on 2018-08-14T13:01:02Z (GMT). No. of bitstreams: 1 Brissac_RafaeladeMenezesSouza_M.pdf: 916033 bytes, checksum: c1e08731f0a8c984aaf530ae5d44f431 (MD5) Previous issue date: 2009 / Resumo: A evasão de estudantes do ensino superior evidencia-se como fenômeno complexo, com conseqüências de ordem pessoal, institucional e social. Entre os estudos que buscam aprofundar o conhecimento sobre este fenômeno encontram-se os que se detém a pesquisar as causas e fatores associados à evasão de alunos neste nível de ensino. No que se refere às variáveis associadas à ocorrência da evasão, os estudos têm mostrado que é possível localizar fatores que são anteriores ao ingresso do estudante no ensino superior e fatores decorrentes da vivência acadêmica durante o curso superior. Sobre os fatores anteriores ao ingresso tem sido sugerido, por pesquisadores do tema, a divisão em três grandes categorias: o background familiar, as experiências educacionais e os atributos individuais gerais. A identificação das variáveis anteriores ao ingresso associadas à evasão fornece elementos para a criação de programas de intervenção que visem à diminuição da ocorrência do fenômeno e que, por sua vez, atuem já sobre o primeiro ano do estudante no ensino superior, considerado como o período crítico em que ocorre o maior número de evasões. A ampla compreensão deste processo demanda a necessidade de expandir os estudos para as diferentes modalidades de cursos de graduação. Esta pesquisa teve dois objetivos: 1) a identificação das variáveis anteriores ao ingresso preditivas de evasão dos anos iniciais, em estudantes de Cursos Superiores de Tecnologia; 2) a identificação das características anteriores ao ingresso preditivas de evasão, em diferentes modalidades (abandono, cancelamento de matrícula, desligamento de ingressante e migração interna). O estudo foi desenvolvido com dados de estudantes de um Centro Superior de Educação Tecnológica, que pertence a uma universidade pública do interior do estado de São Paulo. Na análise proposta foram utilizados dados provenientes do questionário sociocultural, respondido pelos estudantes no momento da inscrição para o vestibular, dados fornecidos pela diretoria acadêmica desta instituição sobre a condição do aluno após o segundo ano de curso (matriculado ou evadido) e modalidade de evasão. Foram analisados, através da Regressão Logística, os dados de 546 estudantes evadidos e não-evadidos nos dois anos iniciais de curso, que ingressaram na instituição no período compreendido entre 2000 e 2004. Os resultados indicaram que a nota de matemática no exame vestibular e o grau de decisão do estudante quanto a escolha de curso são as características anteriores ao ingresso, com maior força preditiva de evasão no início de Cursos Superiores de Tecnologia. No que se refere às modalidades de evasão, observou-se que diferentes características anteriores ao ingresso se mostraram associadas às diferentes modalidades. A identificação de variáveis preditivas, que sinalizem grupos que possam ser mais vulneráveis à interrupção da graduação no momento inicial do curso, sugere a possibilidade de realização de novos estudos, com o intuito de aprofundar sobre qual tipo de programa de intervenção pode ser mais eficaz na redução dos índices de evasão no ensino superior. / Abstract: The dropout of college students has been evident as a complex phenomenon, with consequences of personal, institutional and social nature. Studies about this phenomenon point factors related to the college students dropout. As for the variable associates to the occurrence of the evasion, the studies have shown that it is possible to locate factors that are previous to the student's admission and decurrent factors of the academic experience during the superior course. About factors before the admission, researchers suggest a division in three main categories: the family background, the educational experiences and the student's individual attributes. The identification of variables before the admission connected to the dropout provides elements for the creation intervention programs in order to minimize the occurrence of this phenomenon. These programs would be put into action on the student's first year in college (that is considered the most critical period) when most of the dropout occur. A wider comprehension of this process requires the expansion of the studies on different types of graduation courses. This research had two objectives: 1) the identification of previous admission variables that could predict dropout in the early years among technology college students; 2) the identification of characteristics that were previous to the admission that could predict dropout in different modalities (abandonment, enrollment cancellation, freshman disconnection and course change). The study was developed with data of students from a Technology Education Center, which belongs to a public university located in the countryside of São Paulo state. In the analysis, it was used data from a socialcultural survey, answered by students at the moment of their application for the university admission exam, data provided by the university about the students' situation after the second year in college (enrolled students or students who quit college) and type of dropout. It was analyzed, through Logistic Regression, data from 546 students, who quit college in the first two years of the course, who were admitted between 2000-2004. The results point that the Mathematics grade in the university admission exam and the student decision level about the choice of the course are the strongest predicting dropout factors in the beginning of the technology college courses. About the types of dropout, it was pointed that different previous admission characteristics were connected to the different modalities. The identification of predicting variables pointing those groups that could be more vulnerable to the initial year's course dropout suggests new studies that identify efficient interventions and actions in order to minimize this occurrence. / Mestrado / Psicologia Educacional / Mestre em Educação

Direct marketing problems have been intensively reviewed in the marketing literature recently, such as purchase frequency and time, sales profit, and brand choices. However, modeling the customer response, which is an important issue in direct marketing research, remains a significant challenge. This thesis is an empirical study of predicting customer response to direct marketing and applies a Bayesian approach, including the Bayesian Binary Regression (BBR) and the Hierarchical Bayes (HB). Other classical methods, such as Logistic Regression and Latent Class Analysis (LCA), have been conducted for the purpose of comparison. The results of comparing the performance of all these techniques suggest that the Bayesian methods are more appropriate in predicting direct marketing customer responses. Specifically, when customers are analyzed as a whole group, the Bayesian Binary Regression (BBR) has greater predictive accuracy than Logistic Regression. When we consider customer heterogeneity, the Hierarchical Bayes (HB) models, which use demographic and geographic variables for clustering, do not match the performance of Latent Class Analysis (LCA). Further analyses indicate that when latent variables are used for clustering, the Hierarchical Bayes (HB) approach has the highest predictive accuracy.

Thesis (M.Sc. (Statistics)) -- University of Limpopo, 2015 / Introduction: The high number of variables from the 2009 General Household Survey is prohibitive to do holistic analysis of data due to high correlations that exist among many variables, making it virtually impractical to apply traditional methods such as multinomial logistic regression. The purpose of this study to identify observed variables that can be explained by a few unobservable quantities called factors, using factor analysis. Methods: Factor analysis is used to describe covariance relationships among 162 variables of interest in the 2009 General Household Survey (GHS) and 2009 Quarterly Labour Force Survey of South Africa (QLFS). Data for the respondents aged 15 years and above was analysed by first applying factor analysis to the 162 variables to produce factor scores and develop models for five core areas: education, health, housing, labour force and social development. Multinomial logistic regression was then used to model educational levels and service satisfaction using identified factor sores. Results: The variability among the 162 variables of interest was described by only 29 factors identified using factor analysis, even though these factors are not measured directly. Multinomial logistic regression (MLR) analysis showed negative and significant impact of education factors (fees too high, violence and absence of parental care) on levels of educational attainment. “Historically advantaged” factor is the only factor significant and positively affects educational levels. Housing and social development factors were regressed against service satisfaction. Housing factors such as the home owners, age of a house and male household heads were found to be significant. Social development factors such as “no problem with health”, sufficient water, high income, household size and telephone access were found to be significant. Labour force factors such as employment, industrial business and occupation, employment history and long-term unemployment have positive and significant impact on levels of education. Conclusion: It can be concluded that factor analysis as a data reduction technique has managed to describe the variability among the 162 variables in terms of just 29 unobservable variables. Using MLR in subsequent analysis, this study has managed to identify factors positively or negatively associated with educational levels and service satisfaction. The study suggests that educational, housing, social development and labour force facilities should be improved and education should be used to improve life circumstances. Keywords: factor analysis, factors, multinomial logistic regression, logits, educational levels of attainment, service satisfaction, quality of service delivery. / DST-NRF, Centre of Excellence in Mathematical and Statistical Sciences (MaSS)