Research Article
ADES: A New Ensemble Diversity-Based Approach for Handling Concept Drift
Table 1
Prediction accuracy and CPU time of the four algorithms on both artificial and real-world datasets.
| | Dataset | ADES | DDD | Dynse | ESDE | | Accuracy | CPU-time | Accuracy | CPU-time | Accuracy | CPU-time | Accuracy | CPU-time |
| | Hyperplane | 92.12 (1) | 33.2 (1) | 74.27 (3) | 39.6 (3) | 81.6 (2) | 36.48 (2) | 71.33 (4) | 48.34 (4) | | Stagger | 84.53 (2) | 48.53 (1) | 66.85 (3) | 52.48 (3) | 86.47 (1) | 50.32 (2) | 64.89 (4) | 64.53 (4) | | LED | 79.34 (1) | 63.27 (2) | 75.38 (2) | 53.16 (1) | 71.29 (3) | 49.34 (1) | 68.43 (4) | 128.45 (4) | | Waveform | 81.48 (3) | 83.36 (2) | 83.47 (1) | 113.32 (3) | 84.34 (2) | 78.23 (1) | 62.85 (4) | 124.54 (4) | | Covertype | 87.38 (2) | 73.39 (1) | 81.43 (3) | 124.63 (4) | 89.39 (1) | 93.47 (2) | 71.23 (4) | 85.67 (3) | | Airlines | 79.63 (1) | 54.34 (2) | 76.57 (2) | 52.34 (1) | 74.73 (3) | 83.34 (4) | 67.43 (4) | 78.45 (3) | | Amazon | 89.37 (1) | 47.53 (1) | 76.89 (3) | 64.39 (3) | 84.63 (2) | 57.46 (2) | 72.34 (4) | 69.48 (4) | | Pokerhand | 79.84 (1) | 39.56 (2) | 74.67 (3) | 37.28 (1) | 76.43 (2) | 73.48 (4) | 70.37 (4) | 67.48 (3) | | Average rank | 1.5 | 1.5 | 2.5 | 2.38 | 2.0 | 2.5 | 4.0 | 3.38 |
|
|
