Table 2.
Comparison of ESEA With Other ML Models
| Model | Set | Sp | Sn | PPV | NPV | Technique |
|---|---|---|---|---|---|---|
| ESEA | Training | 90.7 | 91.1 | 98.9 | 52.7 | ESEA |
| Validation | 96.2 | 92.8 | 99.4 | 67.6 | ||
| LDA | Training | 44.4 | 96.2 | 94.1 | 55.8 | LDA |
| Validation | 50.0 | 95.1 | 92.4 | 61.9 | ||
MLP 7:7-9-1:1 
|
Training | 92.3 | 70.3 | 98.9 | 25.3 | BP100, CG20, CG0b |
| Validation | 100 | 65.1 | 100 | 30.9 | ||
RBF 4:4-9-1:1 
|
Training | 90.7 | 66.1 | 98.5 | 22.5 | KM, KN, PI |
| Validation | 92.3 | 66.3 | 98.2 | 30.0 | ||
LNN 7:7-1:1 
|
Training | 92.6 | 71.2 | 98.8 | 26.1 | PI |
| Validation | 100 | 66.9 | 100 | 32.1 |
Green indicates positive input. Red indicates negative input.
BP = backpropagation; CG = conjugated gradient; ESEA = Excel Solver Evolutionary algorithm; KM = K-means; KN = K-nearest neighbor; LDA = linear discriminant analysis; LNN = linear neural network; ML = machine learning; MLP = multilayer perceptron; NPV = negative predictive value; PI = pseudoinversion; PPV = positive predictive value; RBF = radial basis function; Sn = sensitivity; Sp = specificity.