Konstanz, Univ., Diplomarbeit, 2006.

Master-Arbeit Univ. St. Gallen, 2008.

Ziel dieser Arbeit ist die Untersuchung der Bedeutung der Spezifikation für Ratingmodelle zur Prognose von Kreditausfallwahrscheinlichkeiten. Ausgehend von dem in der Bankenpraxis etablierten Logit-Modell werden verschiedene Modellerweiterungen diskutiert und hinsichtlich ihrer Eigenschaften als Ratingmodelle empirisch und simulationsbasiert untersucht. Die Interpretierbarkeit und die Prognosegüte der Modelle werden dabei gleichermaßen berücksichtigt. Besonderes Augenmerk wird auf Mixed Logit-Modelle zur Abbildung individueller Heterogenität gelegt. Die Ergebnisse zeigen, dass die Spezifikation einen wichtigen Einfluss auf die Eigenschaften von Ratingmodellen hat und dass insbesondere mit Hilfe von Mixed Logit-Ansätzen sinnvoll interpretierbare Ratingmodelle mit guten Prognoseeigenschaften erlangt werden können. / This thesis aims at analyzing the impact of model specification on credit rating models as used by banks to forecast credit default probabilities. The well-established logit model is extended in various directions. Implications are discussed empirically and with the help of Monte Carlo studies. In this respect, interpretability and forecasting quality are considered simultaneously. A main focus is on mixed logit models that allow for modeling individual heterogeneity. Results indicate that proper model specification has a strong impact on the quality of rating models. In particular it is emphasized that mixed logit models offer well interpretable credit rating models with high forecasting power.

Due to enormous growth rates during the last couple of years CarSharing has become an interesting field of research. This Master Thesis focuses on the analysis of CarSharing users. For this purpose more than 37.000 participants of the "Große SrV-Vergleichsstädte"-Survey were taken into account. After a short literature review and some general information about the statistical methods, the thesis tries to work out some major characteristics of the typical CarSharing user. It will be shown, that a high income and a young age are of central importance. Also when analysing other variables different user quotas are observed, but can mostly be explained with different age or income levels. At the end a binary logit model is sugested to differ between users and non users, by using the previous analysed observed characteristics as input.

Techn. Hochsch., Diss.--Aachen, 2007.

Universiẗat, Diss., 2004--München.

Saarbrücken, Univ., Diss., 2004.

No description available.

The thesis proposes a method for automated model selection using a logistic regression model in the context of prospectivity modeling, i.e. the exploration of minearlisations. This kind of data is characterized by a rare positive event and a large dataset. We adapted and combined the two statistical measures Wald statistic and Bayes' information criterion making it suitable for the processing of large data and a high number of variables that emerge in the nonlinear setting of logistic regression. The obtained models of our suggested method are parsimonious allowing for an interpretation and information gain. The advantages of our method are shown by comparing it to another model selection method and to arti cial neural networks on several datasets. Furthermore we introduced a possibility to induce spatial dependencies which are important in such geological settings.

Due to enormous growth rates during the last couple of years CarSharing has become an interesting field of research. This Master Thesis focuses on the analysis of CarSharing users. For this purpose more than 37.000 participants of the "Große SrV-Vergleichsstädte"-Survey were taken into account. After a short literature review and some general information about the statistical methods, the thesis tries to work out some major characteristics of the typical CarSharing user. It will be shown, that a high income and a young age are of central importance. Also when analysing other variables different user quotas are observed, but can mostly be explained with different age or income levels. At the end a binary logit model is sugested to differ between users and non users, by using the previous analysed observed characteristics as input.:Inhalt Tabellenverzeichnis iii Abbildungsverzeichnis v 1 Einleitung 1 2 Literaturüberblick 3 3 Methodische Grundlagen 7 3.1 Korrigierter Kontingenzkoeffizient nach Karl Pearson 7 3.2 Kolmogorov-Smirnov-Test 8 3.3 Logistische Regression 9 4 Vorbetrachtungen 12 4.1 Datenaufbereitung 12 4.2 Vergleich von tatsächlichen Nutzern und angemeldeten Personen 13 5 Der Typische CarSharing-Nutzer 17 5.1 Alter 17 5.2 Einkommen 22 5.3 Haushaltsgröße 27 5.4 PKW-Verfügbarkeit 36 5.6 Höchste Berufsausbildung 42 5.7 Zusammenfassung 45 6 Logistische Regression 46 7 Fazit 53 Anhang I Literaturverzeichnis III Ehrenwörtliche Erklärung V