In our master thesis, we compare ten classification algorithms for credit scor- ing. Their prediction performances are measured by six different classification performance measurements. We use a unique P2P lending data set with more than 200,000 records and 23 variables for our classifiers comparison. This data set comes from Lending Club, the biggest P2P lending platform in the United States. Logistic regression, Artificial neural network, and Linear discriminant analysis are the best three classifiers according to our results. Random forest ranks as the fifth best classifier. On the other hand, Classification and regression tree and k-Nearest neighbors are ranked as the worse classifiers in our ranking. 1

Master of Science in Veterinary Biomedical Science / Department of Clinical Sciences / Robert L. Larson / The selection of young replacement animals can have a significant impact on beef herd reproductive performance. Replacement heifers can be utilized to improve reproductive performance by replacing mature animals that failed to meet the production with young, cycling heifers that can have the potential of improving the reproductive momentum of a herd. The use of yearling bulls in natural breeding herds has the advantage of shortening the generational interval of the herd and has the potential of reducing the cost per cow exposed as additions to the bull battery. This thesis involves two studies that investigated methods used for the selection of peripubertal replacement animals in beef herds. The first study evaluated the ability of the novel Ready-Intermediate-Problem (RIP) replacement heifer evaluation matrix to classify heifers into groups that allow producers to select for replacements that meet production goals. Beef heifers (n=341) were classified according to the RIP matrix guidelines and then exposed to AI breeding, bull breeding, or a combination of both as per the management plans for each participating herd. Following breeding season the heifers were evaluated to determine pregnancy status, pregnancy status to single AI exposure, days bred, and the number of 21 day cycles needed during breeding season to become pregnant. After breeding season, 298 (87%) of the heifers were pregnant, 204 (68%) of which became pregnant in the first 21 days of the breeding season. Probability of overall pregnancy and pregnancy after single AI exposure was not significantly associated with RIP classification. There was a significant interaction in RIP classification by 21 day cycle. The second study was a retrospective study using BSE result data to determine the proportion of yearling beef bulls that are classified as satisfactory potential breeders when reevaluated after failing their initial breeding soundness evaluation (BSE) and to identify any predictive factors at initial BSE for satisfactory performance at revaluation. The study included 2,805 beef bulls between 11 and 14 months of age at first BSE evaluated at KABSU from 2006 to 2014. Generalized linear mixed models were created to assess potential associations among breed, age, and interaction between breed and age and passing the initial evaluation and identify predictive factors for risk of passing BSE after initial failure. The majority (93%) of the study bulls passed one of up to three BSEs. There was a significant interaction between age and breed of bull at initial BSE. Identification of suitable peripubertal replacement animals that will improve herd reproductive performance remains a challenge for producers. There are several factors that can affect replacement animals’ ability to perform according to expectations at the beginning of the breeding season. Classification of heifers into categories that can predict performance during breeding season with reasonable confidence can assist producers in identifying heifers that complement the reproductive performance goals of the herd. Utilizing BSE to identify bulls that have adequate semen quality as well as other traits important for breeding soundness is similarly important in reducing the risks of using young bulls for breeding.

The social diversity of the South African population holds considerable challenge for psychologists, especially in respect of differences in language, culture and socio-economic context. The implications of the diverse nature of expectations and needs of unique individuals, clients as well as professionals, in particular in psychological assessment, are of concern. Projected storytelling in assessment is widely recognised as valuable, especially when working with children and adolescents. The technique has its pitfalls, including the way the stories produced may be influenced by leading questions, applying different methods of interpretation, and administering the instrument in cross-cultural assessment situations. Psychologists presenting projection plates to adolescent clients in South Africa frequently obtain little more than one-liners from standard procedures, raising doubts about viability and reliability of the technique. Prompting and probing need to be enhanced without compromising the projective value of responses or the uniqueness of clients. Feuerstein pioneered mediated intervention for learners with cognitive barriers, and the dynamic assessment of culturally different children. In this study, a dynamic assessment technique of questioning (DATQ) was used to actualise projection potential in mono- and cross-cultural assessment situations. The aim of the study was two-fold: to investigate the influence of a DATQ with projection plates during the psychological assessment of adolescents, and to investigate the influence of culture on such assessment in mono- and cross-cultural situations. A qualitative, multiple case study of ten participants representing five language groups in South Africa was undertaken within a predominantly postmodern epistemology. The tension between assessment from the positivist and post-modern paradigms was acknowledged through applying different perspectives during different stages of the research. A test-training-continuation-of-test situation was created for the administration of seven projection pictures, after which two discussion protocols were used. Data-analysis and interpretation took place in four phases by way of projection analysis (using the Bellak TAT Analysis Blank and Haworth’s analysis of defences), structural analysis (with categories such as word-count, response pattern, formulation, number of statements, prompts, hesitations, repetitions), analysis of the participant’s experience of the Murray-method versus the dynamic assessment technique of questioning, and analysis of possible cross-cultural influences on the assessment (utilizing the Scoring Sheet for the Psychocultural Scoring System (SSPSS) and triangulating the results with the projection analysis and the thematic analysis of the conversation about culture). Findings were derived from intra-comparison (per participant) and inter-comparison (per phase of the assessment) of the analyses. The main conclusions of the study point towards participants’ projective responses increasing and deepening in the direction of self-understanding and wholesome problem solution as well as being structurally enhanced, their emotional experience of the assessment situation being positive, culturally associated values being expressed and cultural barriers to interaction being lessened in both mono- and cross-cultural assessment. Whilst projection isn’t an exclusively context-bound phenomenon and generally occurs irrespective of cultural specificity, it was found that supporting clients through non-directive prompting to voice their associations apparently didn’t interfere with the unconscious content being solicited, irrespective of the mono-/cross-cultural nature of the assessment. / Thesis (PhD)--University of Pretoria, 2011. / Educational Psychology / unrestricted

First degree relatives (FDRs) of subjects with early onset of coronary heart disease (CHD) have higher risk of developing cardiovascular disease. We verified early CHD by angiography in the index patients and extensively phenotyped their FDRs to investigate the relationship of traditional and non-traditional cardiovascular risk factors to carotid ultrasound and coronary calcium scoring findings. B-mode carotid ultrasound was used to assess the combined intima-media thickness and plaque burden in 111 FDRs. The biochemical and anthropometrical characteristics of the FDRs were compared with those of healthy controls matched for sex, age, ethnicity and BMI. Odds ratios indicate that FDRs are more likely to have positive carotid ultrasound findings compared to controls; 2.23 (95% CI 1.14 – 4.37) for intima-media thickness and 2.3 (95% CI 1.22 - 4.35) for average total thickness. In multivariate analysis positive carotid ultrasound findings were higher in FDRs independent of age, gender, total cholesterol over HDL-c ratio, systolic blood pressure and smoking but not homocysteine which had higher values in FDRs compared to controls. In conclusion FDRs of patients with angiographically confirmed CHD have higher burden of subclinical atherosclerosis even when considered in the context of traditional risk factors. Coronary artery calcium scoring (CAC), assessed by 64-slice multi-detector computed tomography (MDCT), was used to assess burden of subclinical atherosclerosis in 57 FDRs compared to controls. FDRs have a two-fold increase in risk of having CAC positive findings; odds ratios for the 75th percentile was 1.96 (95%CI 1.04 – 3.67, p<0.05) while for the 90th percentile odds ratio was 2.59 (95% 1.232 – 5.473, p<0.05). In summary, the risk of significant CAC findings, measured by 64-slice MDCT, is two-fold higher in FDRs than controls. These findings correlate highly with carotid ultrasound findings in the same cohort. Different thresholds for CAC may be appropriate when assessing male versus female FDRs. Together increased carotid ultrasound findings and CAC scoring results in FDRs of patients with validated early onset of CHD suggest these imaging techniques as potentially useful tools in cardiovascular risk assessment that will go above and beyond the current diagnostic algorithms. / Medicine, Faculty of / Pathology and Laboratory Medicine, Department of / Graduate

This thesis takes up the issue of consumer loans credit risk. It aims to identify factors that influence the likelihood of default during the repayment of bank retail products. Theoretical part is focused on the theory of bank loans, credit risk together with its regulation within the Basel guidelines and a description of the most commonly used statistical methods for the creation of scoring models. In the practical part, logistic regression parameter estimates are used on real set, the main features linking loans in default are examined and their impact on the proportion of bad loans in the portfolio is measured.

Thesis is providing an overview of available theoretical and practical piece of knowledge related to Credit risk management in a non-banking international organization. It focuses on international aspects of the business, underlying risk, its determination and mitigation. Described methods are then applied to real-life risk assessment of particular customer.

The diploma thesis deals with a financial analysis of T - Mobile Czech Republic, a.s. Trying to determine whether the analyzed enterprise is thriving company or company in financial distress. The analysis uses ratios indicators, differential indicators, analysis of financial statements and the scoring and bankruptcy models, economic value added and economic standard as well. Finally, it assessed the situation analyzed company in the context of competition - the mobile operators in the Czech Republic.

The aim of the present work is to describe the application of the logistic regression model to the field of probability of default modeling, and provide a brief introduction to the scoring development process used in financial practice. We start by introducing the theoretical background of the logistic regression model; followed by a consequent derivation of three most common scoring models. Then we present a formal definition of the Gini coefficient as a diversification power measure and derive the Somers-type formulas for its estimation. Finally, the key part of this work gives an overview of the whole scoring development process illustrated on the examples of real business data.

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.

The Rey-Osterrieth Complex Figure Task (ROCFT) has been a standard in neuropsychological assessment for six decades. Many researchers have contributed administration procedures, additional scoring systems and normative data to improve its utility. Despite the abundance of research, the original 36-point scoring system still reigns among clinicians despite documented problems with ceiling and floor effects and poor discrimination between levels of impairment. This study is an attempt to provide a new method based upon item response theory that will allow clinicians to better describe the impairment levels of their patients. Through estimation of item characteristic curves, underlying traits can be estimated while taking into account varying levels of difficulty and discrimination within the set of individual items. The ultimate goal of the current research is identification of a subset of ROCFT items that can be examined in addition to total scores to provide an extra level of information for clinicians, particularly when they are faced with a need to discriminate severely and mildly impaired patients.