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

Table 18. Actual classification performance for TFTF dataset using NBC classifier.

Sensitivity Specificity GMean PPV F1-measure
DWFS 84.7% (±2.79%) 36.91% (±4.4%) 55.82% (±3.24%) 57.59% (±4.43%) 68.46% (±3.2%)
MRMR+DWFS 74.99% (±4.17%) 57.44% (±6.27%) 65.49% (±3.14%) 64.22% (±1.47%) 69.11% (±1.15%)
JMI+DWFS 76.76% (±4.84%) 48.09% (±3.91%) 60.69% (±3.17%) 59.91% (±3.64%) 67.22% (±3.27%)
MRMR 91.2% (±3.97%) 24.09% (±7.09%) 46.4% (±6.83%) 54.89% (±4.63%) 68.41% (±3.72%)
JMI 88.03% (±2.76%) 21.09% (±6.9%) 42.62% (±6.7%) 53.07% (±4.18%) 66.09% (±2.82%)
WEKA 77.47% (±7.48%) 57.7% (±6.58%) 66.6% (±2.99%) 64.95% (±4.85%) 70.49% (±4.45%)
FST3 81.18% (±3.68%) 37.24% (±10.25%) 54.47% (±6.35%) 56.88% (±5.41%) 66.68% (±3.03%)
ALL Features 90.08% (±3.98%) 26.16% (±8.03%) 48% (±7.45%) 55.3% (±4.51%) 68.39% (±3.23%)
Correlation-baseline 90.01% (±4.99%) 25.11% (±10.71%) 46.34% (±10.08%) 54.99% (±5.4%) 68.07% (±4.06%)