Publikationsserver der RWTH Aachen University
Monitoring-Datenbank-Konzepte für Kreditinstitute in der Dritten Welt am Beispiel Togo
Abstract
dc:descriptionTraceability and accountability of financial flows are among the main issues faced by credit institutes and long-term aid measures in crisis regions of the so-called Third World, especially in the context of credit rating and credit evaluation. Databases, data warehouses and business intelligence naturally play a key role in such situations, but equally important are the cultural adequacy and acceptance of solutions in the regional population, business, and administration. The thesis therefore pursues the key research question how concepts of data warehousing and dataming – originally created in the US and Europe – ca be adapted to the needs of banks and other credit organizations in the Third World, more specifically, in the West-African Franc Zone. Based on a case study in Togo, the problems of credit handling in this region as well as the situation concerning data management and data quality are explored. Subsequently, the thesis evaluates solution concepts from the area of data warehousing, specifically metadata management, semantic modeling, and data quality management, with respect to the given context. Finally, an example data set gained from the case study is used to compare a number of data mining method variants and combinations from the Neural Network area with respect to the question which of them might be best suitable for the situation, where under a given risk acceptance of non-repayment (error of type alpha), you want to reduce the number of credit requests which are rejected wrongly (error of type beta) in order to avoid stifling the economy. Among the investigated variants of basic backpropagation networks, cascade correlation, resilient propagation, pre-optimization with genetic algorithms, and fuzzy classification, the best results were achieved backpropagation networks that have a relatively high number of hidden neurons. The results from problem analysis, literature analysis, and selected experiments with experimental data indicate some conclusions for promising strategies in data warehousing and data mining that could form a foundation for future large-scale field studies in the region.
Degree
thesis:*- Grantor dc:publisher
- Publikationsserver der RWTH Aachen University
- Year dc:date
- 2008
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Nayo, Kwami Emmanuel
- Contributors dc:contributor
-
- Jarke, Matthias
Subjects
dc:subject × 14Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- ger