Table 9. Results for different workflows of logistic regression classifiers on German Credit Data, with standard deviations over n = 10 runs.

Model	ROC AUC	Sensitivity	Specificity	Accuracy
No Transformation	0.76 (0.043)	0.71 (0.033)	0.73 (0.039)	0.72 (0.038)
No Transformation, SMOTE	0.76 (0.027)	0.72 (0.046)	0.7 (0.041)	0.71 (0.043)
PCA	0.76 (0.031)	0.71 (0.041)	0.71 (0.035)	0.71 (0.037)
PCA, SMOTE	0.76 (0.049)	0.7 (0.046)	0.74 (0.048)	0.73 (0.048)
Mapper, 1 model	0.68 (0.054)	0.66 (0.063)	0.69 (0.057)	0.68 (0.059)
Mapper, SMOTE, 1 model	0.69 (0.068)	0.66 (0.060)	0.72 (0.061)	0.70 (0.061)
Mapper, 2 models, equal weight	0.73 (0.072)	0.71 (0.065)	0.69 (0.068)	0.70 (0.067)
Mapper, SMOTE 2 models, equal weight	0.72 (0.079)	0.71 (0.073)	0.7 (0.066)	0.70 (0.068)
Mapper, 2 models, AUC weight	.72 (0.075)	0.71 (0.068)	0.71 (0.063)	0.71 (0.064)
Mapper, SMOTE 2 models, AUC weight	0.72 (0.078)	0.7 (0.077)	0.67 (0.063)	0.68 (0.067)
Mapper, Node PCA, 1 model	0.74 (0.056)	0.73 (0.063)	0.67 (0.061)	0.69 (0.061)
Mapper, Node PCA, SMOTE, 1 model	0.71 (0.067)	0.69 (0.054)	0.68 (0.055)	0.683 (0.055)
Mapper, Node PCA, 2 models, equal weight	0.74 (0.057)	0.78 (0.063)	0.67 (0.069)	0.703 (0.068)
Mapper, Node PCA, SMOTE 2 models, equal weight	0.72 (0.073)	0.7 (0.076)	0.66 (0.067)	0.67 (0.069)
Mapper, Node PCA, 2 models, AUC weight	0.74 (0.072)	0.75 (0.059)	0.67 (0.063)	0.69 (0.062)
Mapper, Node PCA, SMOTE 2 models, AUC weight	0.74 (0.079)	0.7 (0.062)	0.69 (0.071)	0.69 (0.067)