Skip to main content

Table 7 Classifier performance with filtering on large default payments dataset

From: Improving binary classification using filtering based on k-NN proximity graphs

ClassifierDecision TreeLogistic regressionNaive BayesSVMNeural NetworkRandom ForestDES-LA
Accuracy0.8130.8180.790.8150.7940.8170.835
Sensitivity0.9510.9580.8750.9560.9850.9540.95
Specificity0.3270.3240.4910.3220.1240.3340.365
AUC0.6880.7130.7120.7140.6830.7290.769
Brier Score0.1790.1620.1920.150.170.1610.159