An assessment on loan performance from combined quantitative and qualitative data in xml

Novita Ikasari, Fedja Hadzic

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

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.

Original languageEnglish
Title of host publicationDiscovery Science - 15th International Conference, DS 2012, Proceedings
Pages268-283
Number of pages16
DOIs
Publication statusPublished - 5 Nov 2012
Event15th International Conference on Discovery Science, DS 2012 - Lyon, France
Duration: 29 Oct 201231 Oct 2012

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume7569 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference15th International Conference on Discovery Science, DS 2012
Country/TerritoryFrance
CityLyon
Period29/10/1231/10/12

Keywords

  • DSM approach
  • Loan performance assessment
  • XML mining

Fingerprint

Dive into the research topics of 'An assessment on loan performance from combined quantitative and qualitative data in xml'. Together they form a unique fingerprint.

Cite this