Table 2 Potential features selected using Meta-Heuristics and XAI feature selection methods.
From: Optimized hybrid machine learning framework for early diabetes prediction using electrogastrograms
Total Features | Features selected using meta-heuristics techniques | Features selected using XAI technique | ||
|---|---|---|---|---|
Genetic Algorithm (GA) | Ant Colony Optimization (ACO) | Simulated annealing (SA) | ||
Variance | Y | Y | Y | Y |
Root Mean Square | Y | N | N | N |
Mean absolute value | N | N | N | N |
Waveform length | N | Y | Y | N |
Skewness | Y | Y | N | Y |
Maximum fractal length | Y | Y | N | Y |
Teager Kaiser energy | N | Y | Y | Y |
R(a) = 0.2 | Y | N | Y | Y |
R(a) = 0.4 | Y | N | N | N |
R(a) = 0.6 | N | N | N | N |
R(a) = 0.8 | Y | N | N | N |
R(a) = 0.9 | N | Y | N | N |
T(a) = 0.2 | Y | N | N | N |
T(a) = 0.4 | Y | N | Y | Y |
T(a) = 0.6 | Y | N | Y | Y |
T(a) = 0.8 | Y | N | Y | Y |
T(a) = 0.9 | Y | N | Y | Y |
Spectral Entropy | N | Y | Y | Y |
Mean Frequency | Y | Y | Y | Y |
Median Frequency | Y | Y | Y | Y |
