A Credit Scoring Ensemble Framework using Adaboost and Multi-layer Ensemble Classification

The accuracy of classification plays a crucial role in the financial industry, and an increase of 1% in the accuracy of credit scoring in credit customer selection, risk measurement, etc. would significantly reduce the losses of financial institutions. Sometimes one particular classifier perform better than others for a given dataset, while the performance may worse than other classifiers for other datasets. Many studies have shown that classifier ensemble method is a more effective approach. a multi-level weighted voting classification algorithm based on the combination of classifier ranking and Adaboost algorithm is proposed in this paper.. Four feature selection methods are used to select the features, and then seven commonly used heterogeneous classifiers are used to select five classifiers and calculate their ranks, and then AdaBoost is used to boost the performance of the selected base classifiers and calculate the updated F1 and ranks. The effects of ensemble framework Majority Voting (MV), Weighted Voting (WV), Layered Majority Voting (LMV), Layered Weighted Voting (LWV) were all evaluated from the aspects of accuracy, sensitivity, specificity, and G-measure. In addition, the ROC curves of each ensemble framework are plotted for analysis, and the outcome of the experiments shows that our presented method achieves significant results on Australian credit score data and some progress on the German loan approval data.