Model Number of features Accuracy Sensitivity Specificity AUC KNN 360.74250.36870.88990.7229 SVM 360.83500.40141.00000.8353 LR 360.80500.64210.86800.8464 RF 360.82080.54950.92030.8776 DTC 360.79420.66840.84140.7549 GBDT 360.81420.48870.93310.8730 XGBoost 360.81670.58180.90510.8522
Variables Important score AH 0.0888 L 0.0863 G 0.0635 AG 0.0609 AQ 0.0533 AN 0.0533 AO 0.0533 X 0.0507 F 0.0507 AE 0.0482 BA 0.0457 AB 0.0431 AI 0.0431
同样,我们用上面选出的这13个变量建模,实验设置和之前一样,实验结果如下:
1 2 3 4 5 6 7 8
Model Number of features Accuracy Sensitivity Specificity AUC KNN 130.74670.36710.88250.7229 SVM 130.85500.44571.00000.8419 LR 130.81670.55130.90830.8418 RF 130.85250.61700.93490.8637 DTC 130.82080.70300.85930.7812 GBDT 130.84000.57670.93560.8775 XGBoost 130.83330.58350.92080.8489
Model Number of features Accuracy Sensitivity Specificity AUC KNN 20.76330.47590.87750.8033 SVM 20.76830.40450.91730.7980 LR 20.77000.33830.94410.8321 RF 20.82170.64170.89410.8239 DTC 20.82750.67880.88970.7843 GBDT 20.80080.50130.92280.8517 XGBoost 20.77330.55590.86230.8337
Model Number of features Accuracy Sensitivity Specificity AUC KNN 20.76720.39060.89750.7578 SVM 20.77890.28360.95200.6928 LR 20.77780.21260.97870.7921 RF 20.78390.27810.95990.7801 DTC 20.77890.25300.96230.7633 GBDT 20.78280.19960.98730.7735 XGBoost 20.76750.26790.94200.7764