Research Article
The Impact of Simulated Spectral Noise on Random Forest and Oblique Random Forest Classification Performance
Table 1
RF classification accuracies for three simulated noise levels for three weeks of infestation.
| Noise (%) | Week one | Week two | Week three | 0 | 10 | 20 | 30 | 0 | 10 | 20 | 30 | 0 | 10 | 20 | 30 |
| Train accuracy | 96.19 | 93.88 | 93.22 | 92.50 | 93.34 | 92.39 | 90.18 | 84.07 | 92.73 | 92.08 | 89.12 | 88.28 | Test accuracy | 91.07 | 87.50 | 85.71 | 83.92 | 94.64 | 91.07 | 91.07 | 83.92 | 87.50 | 89.28 | 89.28 | 89.28 | Train kappa | 92.30 | 87.68 | 86.35 | 84.92 | 86.94 | 84.76 | 80.14 | 68.34 | 85.47 | 84.18 | 78.30 | 76.66 | Test kappa | 82.14 | 75.00 | 71.42 | 67.85 | 89.28 | 82.14 | 82.14 | 67.85 | 75.00 | 78.57 | 78.57 | 78.57 |
|
|