Research Article
An Improved Fuzzy Based Missing Value Estimation in DNA Microarray Validated by Gene Ranking
Table 12
Performance comparison for Prostate dataset.
| Dataset name | Prostate | Number of genes | Number of samples | Number of samples in Cancer dataset | Number of samples in normal dataset | Number of samples in training dataset | Number of samples in testing dataset | SVM kernel used | Number of folds for LOOCV |
| 12600 | 102 | 52 | 50 | 82 (42 cancer, 40 normal) | 20 (10 cancer, 10 normal) | Linear | 41 |
| Algorithm | % accuracy | Sensitivity/recall | Specificity | 1-score | -mean |
| Proposed approach | 100 (with top 5 genes) | 1 | 1 | 1 | 1 | PSO based graph theoretic approach | 91 | 0.91 | 0.92 | 0.91 | 0.91 | DWFS using KNN classifier | 86 | 0.87 | 0.85 | 0.86 | 0.86 | DWFS using NBC classifier | 80 | 0.76 | 0.85 | 0.78 | 0.80 |
|
|