Table 5

Performance evaluation of the over-sampling method in terms of sensitivity and specificity
Sensitivity Specificity
# of terms SVM Logistic Ridge SVM Logistic Ridge
10 0.6544 ± 0.026 0.6494 ± 0.027 0.6288 ± 0.020 0.8577 ± 0.012 0.8580 ± 0.012 0.8586 ± 0.015
20 0.4796 ± 0.020 0.5051 ± 0.020 0.4736 ± 0.019 0.9260 ± 0.006 0.9235 ± 0.006 0.9284 ± 0.007
30 0.3487 ± 0.023 0.3741 ± 0.024 0.3643 ± 0.035 0.9484 ± 0.004 0.9447 ± 0.005 0.9265 ± 0.032
40 0.2831 ± 0.017 0.3291 ± 0.018 0.2791 ± 0.026 0.9563 ± 0.004 0.9385 ± 0.004 0.9406 ± 0.023
50 0.2958 ± 0.024 0.3582 ± 0.025 0.2214 ± 0.025 0.9466 ± 0.006 0.9089 ± 0.010 0.9569 ± 0.023

Sparse feature is used and the classification performance on stage range 11-12 is reported. Three different classifiers are applied for comparison, namely, SVM with linear kernel (SVM), logistic regression (Logistic) and ridge regression (Ridge).

Yuan et al.

Yuan et al. BMC Bioinformatics 2012 13:107   doi:10.1186/1471-2105-13-107

Open Data