1 Introduction
2 Proposed updating mechanism
2.1 Correlation coefficient
2.2 Information entropy
2.3 Updating mechanism based on correlation coefficient and entropy
2.3.1 Data processing
2.3.2 The establishment of mathematical model
2.3.3 Updating model
3 Classifier pool
3.1 Shape extraction of data block: convex hull
3.2 The working mechanism of classifier pool
4 Experimental results and analysis
Time | Correlation coefficient | Entropy value | System model |
---|---|---|---|
0, 1 | 0.9284 | 0.0995 | No concept drift |
0, 2 | 0.9089 | 0.1253 | No concept drift |
0, 5 | 0.8755 | 0.1680 | No concept drift |
0, 6 | 0.7081 | 0.3526 | Concept drift, accumulating |
0, 7 | 0.7953 | 0.2627 | Concept drift, accumulating |
0, 8 | 0.5283 | 0.4864 | Concept drift, accumulating |
0, 10 | 0.6445 | 0.4085 | Concept drift, accumulating |
10, 13 | 0.9939 | 0.0087 | No concept drift |
3, 21 | 0.1573 | 0.4197 | Concept drift |