Effects of Pooling Samples on the Performance of Classification Algorithms: A Comparative Study

Figure 2

Statistically significant matrices of classifiers performances among various pool sizes datasets. Statistically significant differences of performances of classifiers among different pool sizes of 1 (individual), 2, 3, and 5 are shown using Wilcoxon rank sum test, respectively. (a) and (b) show the matrices when numbers of synthesized markers are 2 and 10, respectively. Each square represents a comparison between two pool sizes datasets. The colors indicate the level of significance.