Research Article

A New Rule-Based Classification Method Using Shape-Based Properties of Spectral Curves

Table 4

Classification results for real waste dataset.

MetricsRule-based methodSVM (a)SVM (b)SVM (c)SVM (d)SVM (e)

Overall accuracy0.98420.97700.98260.98050.97600.9586
Precision0.98420.93980.96980.98070.98560.9854
Sensitivity0.94980.94770.94340.93530.92080.8831
False positive Rate0.00750.00930.00980.01190.01520.0267
F1-score0.96610.93960.95400.95560.95040.9290
Kappa0.95060.92830.94560.93920.92490.8707

(a) Solver: C-SVC; classification type: one-versus-all; kernel: RBF; C-value: 0.2; threshold: 0.2. (b) Solver: C-SVC; classification type: one-versus-all; kernel: RBF; C-value: 0.2; threshold: 0.3. (c) Solver: C-SVC; classification type: one-versus-all; kernel: RBF; C-value: 0.2; threshold: 0.4. (d) Solver: C-SVC; classification type: one-versus-all; kernel: RBF; C-value: 0.2; threshold: 0.5. (e) Solver: C-SVC; classification type: one-versus-all; kernel: RBF; C-value: 0.2; threshold: 0.6.