Table 4 Classification accuracies and the number of selected genes for the two different mutual information based methodologies (Uncertainity based (UB) and Ding and Peng's (DP)). The accuracies as well as the number of genes selected in iterations 1 and 2 respectively are shown for the UB method while the accuracies and genes selected for two different runs are shown for the DP case. For both methodologies, the accuracies reported are LOOCV accuracies.
From: An Entropy-based gene selection method for cancer classification using microarray data
Algorithm | Colon | Leukemia | NCI | |||
|---|---|---|---|---|---|---|
| Â | % Acc | # Genes | % Acc | # Genes | % Acc | # Genes |
UB (ours) | 90.3/91.9 | 29/9 | 80.6/76.4 | 21/5 | 57.6/52.5 | 59/15 |
DP | 75.8/91.9 | 50/20 | 98.6/100 | 50/10 | 73.3/61.7 | 50/20 |