ARE DECISION TREES THE BEST TOOL FOR IMPROVEMENT OF THE CLASSIFICATION ACCURACY RATES AND EXPLAIN ABILITY OF LOAN GRANTING DECISIONS?

摘要

The paper compares the classification performance rate of eight models: logistic regression (LR), neural network (NN), radial basis function neural network (RBFNN), support vector machine (SVM), case-base reasoning (CBR), and three decision trees (DTs). We build models and test their classification accuracy rates on a historical data set provided by a German financial institution. The data set contains 21 financial attributes of 1000 customers of which 300 defaulted upon a loan and 700 paid it off. To obtain reliable and unbiased error estimates for each of the eight models we apply 10-fold cross-validation and repeat an experiment 10 times. We found that in the overall classification accuracy rates at 0.5 probability cut-off, two of the three DT models significantly outperformed (at a=0.05) the other remaining models. We then concentrate our attention on DT models and compare their performance at 0.3 and 0.7 cut-off levels which are more likely to be used by financial institutions. The DT models not only classify better than the other models, but the knowledge they leam in the form of if-then rules is easy to interpret, makes sense, and may be of value to financial institutions which may have to explain the reasons for a loan denial.