haku: @indexterm consumer credit / yhteensä: 61
viite: 21 / 61
Tekijä: | Hofmann, H.-J. |
Otsikko: | Die Anwendung des CART-Verfahrens zur statistischen Bonittsanalyse von Konsumentenkrediten |
Lehti: | Zeitschrift für Betriebswirtschaft
1990 : SEP, VOL. 60:9, p. 941-962 |
Asiasana: | CREDIT CONSUMER CREDIT LINEAR MODELS CREDIT MANAGEMENT |
Kieli: | ger |
Tiivistelmä: | Traditional systems of creditscoring rely dominantly on linear discriminant analysis as a statistical tool which can only be applied to quantitative variables. Quantitative variables have to be rescaled by credit experts using their subjective experiences which are often not uniform. In the article the CART (classification and regression trees) method developed by Breiman et al. 1984 is applied in the framework of consumer credit granting. It can be applied to mixed data sets which contain qualitative together with quantitative variables without rescaling. It is shown that CART has major advantages over discriminant analysis in practical creditscoring without loosing much of classification performance. |
SCIMA