Knowledge Based Systems Group, University of Twente, NL
The Pareto project of the knowledge based systems group atthe University of Twente, is directed at the development, analysis, and comparison of machine learning methods applied to the domain ofbusiness and finance. More specifically we are interested insupervised learning of classification functions from data.
Here's a number of problems we are either working on atthe moment, or interested in for future research:
- Credit evaluation: currently working on in cooperation with Dutch ABN-AMRO bank.
- Prediction of corporate bankruptcy: research topic of Paul Pompe, a Ph D student of our group.
- Methodological issues in the empirical comparison of several machine learning and statistical methods for classification, a.o. classification trees, neural networks, statistical discriminant analysis. How should we evaluate methods, and how can we make an honest comparison?
- Other interesting applications that we may address in the future: target marketing (distinguish potential buyers from non-buyers), insurance risk assessment, market segmentation. The number of interesting applications only seems to be limited by the imagination.
Some of this research might not qualify as Data Mining(e.g. the corporate bankruptcy research), sincethe data are especially collected for the purpose of learning.
The credit evaluation project uses an existing database of thebank, in which data of clients that received a loan are recorded,together with information on their "pay-back behaviour". This database will be used to learn a rule to distinguish "payers"from "non-payers". This rule will subsequently be used toevaluate future loan applications.
University of Twente
Department of Computer Science
Knowledge Based Systems Group
P.O. Box 217
7500 AE Enschede
Note: This info converted from the original "The Data Mine" pages and pre-dates June 2001. Please remove this note if you update or check the info