Skip to main content
. 2015 Feb 26;10(2):e0117988. doi: 10.1371/journal.pone.0117988

Table 11. Actual classification performance for Medelon dataset using KNN classifier.

Sensitivity Specificity GMean PPV F1-measure
DWFS 77.99% (±2.05) 79.33% (±4.47) 78.63% (±2.58) 78.99% (±5.24) 78.44% (±3.28)
MRMR+DWFS 53.01% (±4.15) 51% (±4.03) 51.9% (±1.88) 51.97% (±3.52) 52.35% (±2.51)
JMI+DWFS 88.53% (±2.18) 89.69% (±1.71) 89.09% (±0.74) 89.56% (±1.73) 89.01% (±1.07)
MRMR 49.16% (±1.3) 54.82% (±2.38) 51.9% (±1.21) 52.12% (±3.74) 50.54% (±1.98)
JMI 76.91% (±2.89) 76.78% (±3.68) 76.79% (±0.71) 76.81% (±3.9) 76.75% (±1.27)
WEKA 87.92% (±2.03) 88.36% (±1.67) 88.12% (±0.45) 88.24% (±2.3) 88.04% (±0.55)
FST3 89.28% (±2.17) 89.79% (±0.69) 89.52% (±0.93) 89.67% (±1.57) 89.45% (±1.05)
ALL Features 68.8% (±4.61) 72.99% (±4.27) 70.79% (±2.43) 71.84% (±4.04) 70.19% (±3.35)
Correlation-baseline 81.18% (±2.27) 82.7% (±1.71) 81.93% (±1.18) 82.41% (±2.09) 81.75% (±0.51)