Skip to main content

Advertisement

Table 6 Comparative results between the model in Do et al. and the proposed model via LASSO

From: A practical approach to Sasang constitutional diagnosis using vocal features

    Male   Female  
    Predicted SC Sensitivity Predicted SC Sensitivity
    TE SE SY Total   TE SE SY Total  
Do et al.             
   TE 185 31 43 259 71.4% 204 34 150 388 52.6%
Train True SC SE 52 62 33 147 42.2% 95 63 144 302 20.9%
SY 84 26 77 187 41.2% 137 43 229 409 56.0%
Total 321 119 153 593   436 140 523 1099  
Accuracy 54.6% 45.1%
   TE 46 6 15 67 68.7% 51 15 60 126 40.5%
Test True SC SE 14 4 16 34 11.8% 32 20 38 90 22.2%
SY 37 12 15 64 23.4% 37 11 43 91 47.3%
Total 97 22 46 165   120 46 141 307  
Accuracy 39.4% 37.1%
Proposed             
   TE 184 22 53 259 71.0% 212 20 156 388 54.6%
Train True SC SE 86 30 31 147 20.4% 119 26 157 302 8.6%
SY 110 15 62 187 33.2% 137 26 246 409 60.1%
Total 380 67 146 593   468 72 559 1099  
Accuracy 46.5% 44.0%
   TE 49 2 16 67 73.1% 70 7 49 126 55.6%
Test True SC SE 16 10 8 34 29.4% 30 7 53 90 7.8%
SY 41 3 20 64 31.3% 43 1 47 91 51.6%
Total 106 15 44 165   143 15 149 307  
Accuracy 47.9% 40.4%