Table 4.
Actual | Output | Accuracy | Sensitivity | Specificity | ||
---|---|---|---|---|---|---|
Cancer | Normal | |||||
M191 | ||||||
Univariate classification | Cancer | 46 | 4 | 87.37% | 92.00% | 82.00% |
Normal | 9 | 41 | ||||
M401 | ||||||
Univariate classification | Cancer | 38 | 12 | 82.46% | 76.00% | 88.00% |
Normal | 6 | 44 | ||||
M311 | ||||||
Univariate classification | Cancer | 40 | 10 | 83.12% | 80.00% | 86.00% |
Normal | 7 | 43 | ||||
M191+M401+M311 (Univariate feature selection + Multivariate classification) | ||||||
Decision Tree (Confidence=0.25, Pruning=true) | 85.01% | 86.00% | 84.00% | |||
Random Forest (Tree=500, Feature=6) | 90.00% | 90.00% | 90.00% | |||
SVM (Gaussian) (Cost=55, Gamma=0.33, # of SVs=17) | 92.27% | 96.00% | 88.00% | |||
Classification results for the feature set from the univariate method (Table 1C) |
Contingency table showing number of cases classified for each of the diagnostic classes for the feature set from the univariate method (Table 1C).