Skip to main content
. 2020 Nov 5;10:19128. doi: 10.1038/s41598-020-76129-8

Table 5.

Comparison of the performance of various combinations of feature selection and classification methods.

Feature selection method Prediction method mse roc_auc roc_auc_prob Accuracy f1_score Precision_score Recall_score Sensitivity Specificity mcc
Mann_Whitney RandomForest 0.088235 0.911765 0.932526 0.911765 0.911458 0.917544 0.911765 0.852941 0.970588 0.829288
Mann_Whitney SVM 0.102941 0.897059 0.943772 0.897059 0.897037 0.897403 0.897059 0.882353 0.911765 0.794461
Boruta RandomForest 0.102941 0.897059 0.933391 0.897059 0.895956 0.914634 0.897059 0.794118 1 0.811503
DCor RandomForest 0.102941 0.897059 0.916955 0.897059 0.896499 0.905836 0.897059 0.823529 0.970588 0.802846
Boruta SVM 0.117647 0.882353 0.947232 0.882353 0.882251 0.883681 0.882353 0.852941 0.911765 0.766032
RFE_RF SVM 0.117647 0.882353 0.947232 0.882353 0.882251 0.883681 0.882353 0.852941 0.911765 0.766032
RandomForest RandomForest 0.117647 0.882353 0.932526 0.882353 0.88143 0.894643 0.882353 0.794118 0.970588 0.776899
DCor LogisticRegression 0.132353 0.867647 0.949827 0.867647 0.865287 0.895349 0.867647 0.735294 1 0.762493
Mann_Whitney LogisticRegression 0.132353 0.867647 0.949827 0.867647 0.865287 0.895349 0.867647 0.735294 1 0.762493
Mann_Whitney Lasso 0.132353 0.867647 0.949827 0.867647 0.865287 0.895349 0.867647 0.735294 1 0.762493
Mann_Whitney ElasticNet 0.132353 0.867647 0.949827 0.867647 0.865287 0.895349 0.867647 0.735294 1 0.762493
DCor Lasso 0.132353 0.867647 0.948962 0.867647 0.865287 0.895349 0.867647 0.735294 1 0.762493
DCor ElasticNet 0.132353 0.867647 0.948097 0.867647 0.865287 0.895349 0.867647 0.735294 1 0.762493
ttest LogisticRegression 0.132353 0.867647 0.933391 0.867647 0.867389 0.870532 0.867647 0.823529 0.911765 0.738173
DCor SVM 0.132353 0.867647 0.933391 0.867647 0.867618 0.867965 0.867647 0.882353 0.852941 0.735612
ttest ElasticNet 0.132353 0.867647 0.930796 0.867647 0.866928 0.875774 0.867647 0.794118 0.941176 0.743376
ttest RandomForest 0.147059 0.852941 0.941176 0.852941 0.852814 0.854167 0.852941 0.882353 0.823529 0.707107
ttest Lasso 0.147059 0.852941 0.934256 0.852941 0.851787 0.864286 0.852941 0.764706 0.941176 0.717137
Boruta LogisticRegression 0.161765 0.838235 0.944637 0.838235 0.835351 0.863721 0.838235 0.705882 0.970588 0.701493
RFE_RF LogisticRegression 0.161765 0.838235 0.944637 0.838235 0.835351 0.863721 0.838235 0.705882 0.970588 0.701493
RandomForest SVM 0.161765 0.838235 0.923875 0.838235 0.836503 0.853207 0.838235 0.735294 0.941176 0.69128
RandomForest Lasso 0.161765 0.838235 0.92301 0.838235 0.833889 0.877778 0.838235 0.676471 1 0.71492
RandomForest ElasticNet 0.161765 0.838235 0.922145 0.838235 0.833889 0.877778 0.838235 0.676471 1 0.71492
RandomForest LogisticRegression 0.161765 0.838235 0.91609 0.838235 0.833889 0.877778 0.838235 0.676471 1 0.71492
Boruta Lasso 0.176471 0.823529 0.943772 0.823529 0.819629 0.854167 0.823529 0.676471 0.970588 0.677003
RFE_RF Lasso 0.176471 0.823529 0.943772 0.823529 0.819629 0.854167 0.823529 0.676471 0.970588 0.677003
Boruta ElasticNet 0.176471 0.823529 0.942042 0.823529 0.819629 0.854167 0.823529 0.676471 0.970588 0.677003
RFE_RF ElasticNet 0.176471 0.823529 0.942042 0.823529 0.819629 0.854167 0.823529 0.676471 0.970588 0.677003
ElasticNet_alpha_.001 LogisticRegression 0.220588 0.779412 0.741349 0.779412 0.771044 0.827254 0.779412 0.588235 0.970588 0.604777
ttest SVM 0.235294 0.764706 0.939446 0.764706 0.757143 0.802372 0.764706 0.941176 0.588235 0.565825
RidgeCV RandomForest 0.235294 0.764706 0.82699 0.764706 0.763889 0.768421 0.764706 0.705882 0.823529 0.533114
RFE_SVM RandomForest 0.235294 0.764706 0.817474 0.764706 0.761404 0.78022 0.764706 0.647059 0.882353 0.544705
ElasticNet_alpha_.001 Lasso 0.235294 0.764706 0.736159 0.764706 0.759505 0.789773 0.764706 0.617647 0.911765 0.553912
ElasticNet_alpha_.001 ElasticNet 0.235294 0.764706 0.734429 0.764706 0.759505 0.789773 0.764706 0.617647 0.911765 0.553912
ElasticNet_alpha_.001 SVM 0.25 0.75 0.762111 0.75 0.748641 0.755526 0.75 0.676471 0.823529 0.505496
SVM LogisticRegression 0.264706 0.735294 0.760381 0.735294 0.71978 0.802222 0.735294 0.5 0.970588 0.533333
RidgeCV SVM 0.279412 0.720588 0.874567 0.720588 0.709989 0.758359 0.720588 0.529412 0.911765 0.477455
RFE_SVM SVM 0.279412 0.720588 0.865052 0.720588 0.696927 0.820755 0.720588 0.441176 1 0.531995
ElasticNet_alpha_.001 RandomForest 0.279412 0.720588 0.75519 0.720588 0.719069 0.725464 0.720588 0.794118 0.647059 0.446026
SVM SVM 0.294118 0.705882 0.776817 0.705882 0.704861 0.708772 0.705882 0.647059 0.764706 0.414644
SVM RandomForest 0.308824 0.691176 0.831315 0.691176 0.658618 0.809091 0.691176 0.382353 1 0.486172
RFE_SVM ElasticNet 0.323529 0.676471 0.851211 0.676471 0.638647 0.803571 0.676471 0.352941 1 0.46291
RFE_SVM Lasso 0.323529 0.676471 0.849481 0.676471 0.638647 0.803571 0.676471 0.352941 1 0.46291
RFE_SVM LogisticRegression 0.323529 0.676471 0.845156 0.676471 0.638647 0.803571 0.676471 0.352941 1 0.46291
SVM Lasso 0.323529 0.676471 0.763841 0.676471 0.645833 0.769841 0.676471 0.382353 0.970588 0.436436
SVM ElasticNet 0.323529 0.676471 0.763841 0.676471 0.645833 0.769841 0.676471 0.382353 0.970588 0.436436
RidgeCV Lasso 0.367647 0.632353 0.885813 0.632353 0.58486 0.744019 0.632353 0.294118 0.970588 0.359425
Lasso_alpha_.001 LogisticRegression 0.367647 0.632353 0.627163 0.632353 0.631636 0.633391 0.632353 0.588235 0.676471 0.265742
RidgeCV ElasticNet 0.382353 0.617647 0.884948 0.617647 0.563241 0.734483 0.617647 0.264706 0.970588 0.332182
RidgeCV LogisticRegression 0.397059 0.602941 0.865917 0.602941 0.540885 0.724105 0.602941 0.235294 0.970588 0.303774
ElasticNet_alpha_.01 RandomForest 0.397059 0.602941 0.676471 0.602941 0.587879 0.620567 0.602941 0.794118 0.411765 0.222812
Lasso_alpha_.001 Lasso 0.426471 0.573529 0.605536 0.573529 0.573437 0.573593 0.573529 0.588235 0.558824 0.147122
Lasso_alpha_.001 ElasticNet 0.426471 0.573529 0.602941 0.573529 0.573437 0.573593 0.573529 0.588235 0.558824 0.147122
Lasso_alpha_.001 RandomForest 0.470588 0.529412 0.663495 0.529412 0.484848 0.544974 0.529412 0.823529 0.235294 0.072739
ElasticNet_alpha_.01 SVM 0.5 0.5 0.545848 0.5 0.333333 0.25 0.5 1 0 0
Lasso_alpha_.01 RandomForest 0.514706 0.485294 0.474048 0.485294 0.326733 0.246269 0.485294 0.970588 0 − 0.12217
ElasticNet_alpha_.01 Lasso 0.558824 0.441176 0.562284 0.441176 0.416441 0.429167 0.441176 0.647059 0.235294 − 0.1291
ElasticNet_alpha_.01 ElasticNet 0.558824 0.441176 0.557958 0.441176 0.416441 0.429167 0.441176 0.647059 0.235294 − 0.1291
Lasso_alpha_.001 SVM 0.573529 0.426471 0.605536 0.426471 0.366005 0.381119 0.426471 0.735294 0.117647 − 0.18699
ElasticNet_alpha_.01 LogisticRegression 0.588235 0.411765 0.553633 0.411765 0.328063 0.324138 0.411765 0.764706 0.058824 − 0.24914
Lasso_alpha_.01 SVM 0.838235 0.161765 0.110727 0.161765 0.139241 0.122222 0.161765 0.323529 0 − 0.71492
Lasso_alpha_.01 LogisticRegression 0.867647 0.132353 0.122837 0.132353 0.116883 0.104651 0.132353 0.264706 0 − 0.76249
Lasso_alpha_.01 Lasso 0.867647 0.132353 0.119377 0.132353 0.116883 0.104651 0.132353 0.264706 0 − 0.76249
Lasso_alpha_.01 ElasticNet 0.867647 0.132353 0.117647 0.132353 0.116883 0.104651 0.132353 0.264706 0 − 0.76249

For methods that produced a regressive score, such as Lasso and ElasticNet, we chose 0.5 as the split point to make a binary classification prediction.