Table 1.
Method | Matthews correlation coefficient (MCC) | Accuracy | Sensitivity | Specificity | AUC | P-value |
---|---|---|---|---|---|---|
iBCE-EL | 0.454 | 0.729 | 0.716 | 0.739 | 0.782 | – |
GB | 0.446 | 0.725 | 0.712 | 0.735 | 0.773 | 0.051 |
ERT | 0.437 | 0.718 | 0.734 | 0.705 | 0.776 | 0.205 |
RF | 0.434 | 0.718 | 0.717 | 0.719 | 0.770 | 0.051 |
AB | 0.396 | 0.702 | 0.662 | 0.722 | 0.737 | 1.2E−16 |
k-NN | 0.301 | 0.644 | 0.715 | 0.591 | 0.691 | 1.1E−9 |
SVM | 0.295 | 0.634 | 0.634 | 0.602 | 0.696 | <2.2E−16 |
LBtope | 0.330 | 0.667 | 0.660 | 0.672 | 0.730 | – |
The first column represents the methods developed in this study. The columns 2–6 respectively represent the MCC, accuracy, sensitivity, specificity, and AUC value. The last column represents McNemar’s Chi-squared test was used to evaluate the performance between iBCE-EL and other methods. A P value <0.05 was considered to indicate a statistically significant difference between iBCE-EL and the selected method (shown in bold). For comparison, we have also included LBtope (LBtope_variable_nr) cross-validation performance on non-redundant data set.