Table 6 A comparative analysis of the accuracy and computational complexity of the proposed approach versus conventional feature selection methods for insect classification is presented.
Feature Selection | Classifier | Accuracy (%) | Models | Train/Prediction Time (s) |
|---|---|---|---|---|
All+PCA | SVM | 85 | 1200 | 12/0.8 |
RF | 87 | 1500 | 15/1.0 | |
KNN | 83 | 800 | 10/0.6 | |
NB | 80 | 500 | 8/0.5 | |
MI | SVM | 86 | 1100 | 11/0.7 |
RF | 88 | 1400 | 14/0.9 | |
KNN | 84 | 700 | 9/0.5 | |
NB | 81 | 450 | 7/0.4 | |
Fisher | SVM | 84 | 1000 | 10/0.6 |
RF | 85 | 1300 | 13/0.9 | |
KNN | 80 | 650 | 8/0.4 | |
NB | 78 | 400 | 6/0.4 | |
Chi-2 | SVM | 83 | 950 | 9/0.6 |
RF | 86 | 1250 | 12/0.8 | |
KNN | 82 | 600 | 7/0.4 | |
NB | 79 | 420 | 6/0.4 | |
MIC | SVM | 87 | 1000 | 10/0.6 |
RF | 88 | 1350 | 13/0.8 | |
KNN | 83 | 700 | 8/0.5 | |
NB | 80 | 400 | 6/0.4 | |
VarThresh | SVM | 82 | 900 | 9/0.5 |
RF | 84 | 1200 | 12/0.8 | |
KNN | 79 | 550 | 7/0.4 | |
NB | 75 | 380 | 5/0.3 | |
Proposed XAI | SVM | 90 | 850 | 8/0.5 |
RF | 92 | 1000 | 9/0.6 | |
KNN | 87 | 550 | 6/0.4 | |
NB | 85 | 350 | 4/0.3 |
