Skip to main content

Table 3 Machine learning experiment test results of gene expression data

From: A novel gene selection method for gene expression data for the task of cancer type classification

Feature Sel. Feature count Accuracy F-Score Roc-Auc FPR MCC
FPKM all features 60483 99.46 ±0.42 99.46 ±0.42 99.45 ±0.50 0.40 ±0.31 99.09 ±0.71
SelectKBest 10000 99.57 ±0.13 99.57 ±0.13 99.61 ±0.18 0.29 ±0.13 99.27 ±0.22
SelectKBest 5000 99.30 ±0.27 99.30 ±0.27 99.47 ±0.22 0.46 ±0.24 98.82 ±0.46
SelectKBest 1000 98.92 ±0.38 98.92 ±0.38 99.13 ±0.41 0.71 ±0.32 98.18 ±0.64
mRMR 5000 98.98 ±0.43 98.98 ±0.43 99.08 ±0.33 0.73 ±0.23 98.28 ±0.72
mRMR 1000 98.44 ±0.52 98.45 ±0.52 98.63 ±0.56 0.82 ±0.32 97.37 ±0.88
Relief 10000 98.44 ±0.31 98.45 ±0.31 98.51 ±0.45 0.82 ±0.28 97.37 ±0.53
Relief 5000 99.30 ±0.36 99.30 ±0.36 99.34 ±0.37 0.27 ±0.14 99.34 ±0.37
Relief 1000 99.46 ±0.17 99.46 ±0.17 99.54 ±0.16 0.39 ±0.14 99.09 ±0.29
Hallmark 4266 99.57 ±0.22 99.57 ±0.21 99.57 ±0.23 0.29 ±0.14 99.27 ±0.36
VCF 3000 Effective Genes 6752 99.57 ±0.47 99.57 ±0.47 99.64 ±0.37 0.31 ±0.34 99.27 ±0.79
VCF 3500 Effective Genes 7741 99.68 ±0.40 99.68 ±0.39 99.72 ±0.30 0.24 ±0.26 99.46 ±0.67