Research Article
Cost-Sensitive Classification for Evolving Data Streams with Concept Drift and Class Imbalance
Table 2
Average G-mean of comparing algorithms.
| | VFDT | AUE2 | KUE | CSDS |
| HyperPlane | 0.93 ± 0.04 (4) | 0.95 ± 0.06 (3) | 0.96 ± 0.05 (2) | 0.97 ± 0.01 (1) | SEA | 0.72 ± 0.10 (4) | 0.83 ± 0.12 (3) | 0.86 ± 0.22 (2) | 0.88 ± 0.01 (1) | LED | 0.82 ± 0.02 (4) | 0.87 ± 0.07 (1) | 0.86 ± 0.06 (2) | 0.84 ± 0.02 (3) | Rotating spiral | 0.73 ± 0.10 (4) | 0.79 ± 0.05 (1) | 0.76 ± 0.02 (3) | 0.78 ± 0.01 (2) | Spam | 0.62 ± 0.08 (4) | 0.66 ± 0.02 (3) | 0.80 ± 0.04 (2) | 0.86 ± 0.02 (1) | Sensor | 0.76 ± 0.11 (3) | 0.74 ± 0.07 (4) | 0.86 ± 0.06 (1) | 0.83 ± 0.04 (2) | Electricity | 0.60 ± 0.05 (4) | 0.64 ± 0.02 (2) | 0.62 ± 0.02 (3) | 0.73 ± 0.02 (1) | Airlines | 0.76 ± 0.07 (3) | 0.74 ± 0.08 (4) | 0.83 ± 0.07 (1) | 0.81 ± 0.03 (2) | Average rank | 3.75 | 2.63 | 2.00 | 1.63 |
|
|