Skip to main content

Table 1 Performance indicators of different classifiers on the breast cancer diagnostic and cardiovascular disease datasets

From: Disease prediction via Bayesian hyperparameter optimization and ensemble learning

ClassifierIndicator
Accuracy (%)Precision (%)Recall (%)F1 score (%)AUCKS Value
XGBoost_BC94.74 (94.40,1.65)92.19 (93.10,3.25)93.65 (91.83,3.68)92.91 (92.38,2.29)0.9857 (0.9845,0.76)0.9061
LightGBM_BC94.74 (94.05,1.69)92.19 (92.65,3.49)93.65 (91.26,3.61)92.91 (92.00,2.33)0.9821 (0.9835,0.80)0.9087
GBDT_BC94.15 (94.72,1.59)90.77 (94.41,3.24)93.65 (92.79,2.21)92.19 (92.96,2.20)0.9856 (0.9869,0.66)0.8968
LR_BC92.40 (93.64,1.62)89.06 (92.77,3.15)90.48 (90.09,3.42)89.76 (91.25,2.19)0.9825 (0.9847,0.58)0.8796
RF_BC92.40 (91.81,1.94)90.32 (90.24,3.64)88.89 (87.77,4.67)89.60 (88.93,2.78)0.9710 (0.9757,0.94)0.8690
BPNN_BC89.47 (92.86,1.85)89.47 (91.64,4.18)80.95 (88.80,3.67)85.00 (90.23,2.56)0.9669 (0.9778,0.92)0.8439
DT_BC87.72 (90.75,1.83)90.38 (86.76,3.84)74.60 (88.53,4.92)81.74 (87.41,2.80)0.9314 (0.9500,1.53)0.6997
XGBoost_CVD73.50 (73.51,0.27)75.80 (75.55.0.49)69.54 (69.53,0.51)72.54 (72.44,0.30)0.8044 (0.8023,0.26)0.4733
LightGBM_CVD73.53 (73.56,0.26)75.38 (75.82,0.47)70.40 (69.17,0.60)72.81 (72.32,0.32)0.8042 (0.8023,0.26)0.4762
GBDT_CVD73.56 (73.51,0.27)75.70 (75.60,0.49)69.90 (69.43,0.58)72.68 (72.38,0.33)0.8041 (0.8023,0.25)0.4746
LR_CVD72.32 (71.92,0.41)74.90 (74.02,0.72)67.69 (67.50,0.54)71.11 (70.62,0.40)0.7869 (0.7829,0.38)0.4503
RF_CVD73.55 (73.51,0.27)75.98 (76.02,0.72)69.39 (68.70,0.60)72.54 (72.17,0.32)0.8026 (0.8012,0.26)0.4717
BPNN_CVD72.85 (72.81,0.31)73.07 (73.73,1.11)72.96 (70.98,1.79)73.01 (72.28,0.53)0.7945 (0.7917,0.28)0.4686
DT_CVD73.26 (73.12,0.17)76.45 (76.79,0.86)67.72 (66.42,1.33)71.83 (71.22,0.48)0.7954 (0.7942,0.30)0.4667
  1. (a) Values in parentheses are the average and standard deviation of the performance indicator values. For the BC dataset, 300 samples were randomly selected from 569 samples each time and repeated 1000 times. For the CVD data set, 1000 samples were randomly selected from 65,535 samples each time and repeated 1000 times.
  2. (b) Italics numbers indicate optimal values