TY - GEN
T1 - An assessment on loan performance from combined quantitative and qualitative data in xml
AU - Ikasari, Novita
AU - Hadzic, Fedja
PY - 2012
Y1 - 2012
N2 - The intensifying need to incorporate knowledge extracted from qualitative information into banks' lending decision has been recognized in recent times, particularly for micro lenders. In this study, the multi-faceted credit information is captured in an integrated form using XML to facilitate the discovery of knowledge models encompassing a broad range of credit risk related aspects. The quantitative and qualitative credit data obtained from the industry partner describes existing lender profiles. The experiments are performed to discover classification models for the performing or non-performing lenders in one problem setting, and the duration of payment delay in another. The results are compared with a common credit risk prediction setting where qualitative data is excluded. The findings confirm the role of domain experts' knowledge as well as qualitative information on loan performance assessment, and describe a number of rules indicating refinement of the banks' lending policy requirement.
AB - The intensifying need to incorporate knowledge extracted from qualitative information into banks' lending decision has been recognized in recent times, particularly for micro lenders. In this study, the multi-faceted credit information is captured in an integrated form using XML to facilitate the discovery of knowledge models encompassing a broad range of credit risk related aspects. The quantitative and qualitative credit data obtained from the industry partner describes existing lender profiles. The experiments are performed to discover classification models for the performing or non-performing lenders in one problem setting, and the duration of payment delay in another. The results are compared with a common credit risk prediction setting where qualitative data is excluded. The findings confirm the role of domain experts' knowledge as well as qualitative information on loan performance assessment, and describe a number of rules indicating refinement of the banks' lending policy requirement.
KW - DSM approach
KW - Loan performance assessment
KW - XML mining
UR - http://www.scopus.com/inward/record.url?scp=84868095047&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-33492-4_22
DO - 10.1007/978-3-642-33492-4_22
M3 - Conference contribution
AN - SCOPUS:84868095047
SN - 9783642334917
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 268
EP - 283
BT - Discovery Science - 15th International Conference, DS 2012, Proceedings
T2 - 15th International Conference on Discovery Science, DS 2012
Y2 - 29 October 2012 through 31 October 2012
ER -