Research Article
ADES: A New Ensemble Diversity-Based Approach for Handling Concept Drift
Table 3
Average accuracy achieved and memory consumption for infinite pool and pruned ADES.
| Benchmark | Infinite pool | Pruned ADES | Accuracy (%) | Memory (MB) | Accuracy (%) | Memory (MB) |
| Hyperplane | 95.6 (5.7) | 16.23 | 92.12 (6.4) | 4.3 | LED | 89.3 (6.8) | 12.08 | 79.34 (10.32) | 8.9 | Waveform | 84.8 (8.3) | 13.24 | 81.48 (8.5) | 10.24 | Stagger | 91.4 (14.6) | 8.9 | 84.53 (5.8) | 5.9 | Airlines | 76.3 (4.3) | 17.14 | 79.63 (12.34) | 14.32 | Covertype | 74.6 (8.5) | 14.17 | 87.38 (14.32) | 12.43 | PokerHand | 73.4 (3.7) | 10.24 | 79.84 (16.21) | 16.34 | Amazon | 78.6 (4.6) | 20.38 | 89.37 (18.34) | 18.26 |
|
|
The numbers in parenthesis indicate the standard deviation between testing batches.
|