Analysis of the Capital Structure of German Companies with the SSA and SVM
Wirtschaftswissenschaftliche Fakultät
This thesis presents and compares the performance of two recently developed classification methods namely the Spatial Stagewise Aggregation procedure and Support Vector Machines. Both techniques are convenient for the application to corporate bankruptcy analysis, in terms of calculation of default probabilities. Repeated random selection simulations varying with respect to variable and record choices for both methods proved a clear superiority in terms of the hit rate, representing the percentage of correctly classified observations, in favor of the SVM. Moreover the thesis presents a way to derive recommendations with respect to the capital structure policy of German manufacturing industry firms on the basis of the evaluated default probabilities.
Dateien zu dieser Publikation