Skip to main content

Table 2 Feature mapping method performance in RBF-epsilon regression SVM modeling, alternatively training and testing between dataset2431 and datatset579 and 10 × cross validation within dataset2431 or datatset579

From: Improving model predictions for RNA interference activities that use support vector machine regression by combining and filtering features

   train2431 train579
   test579 test2431 10 × cross validation test2431 test579 10 × cross validation
Method FN2431
FN579
R MSE R MSE R MSE R MSE
1- 84 0.510 0.095 0.711 0.026 0.485 0.054 0.562 0.079
2- 23 0.379 0.105 0.640 0.029 0.367 0.069 0.500 0.087
3- 23 0.130 0.115 0.094 0.046 0.017 0.138 0.026 0.118
4- 24 0.202 0.115 0.293 0.041 0.214 0.073 0.214 0.041
5- 32 0.214 0.112 0.243 0.042 0.164 0.046 0.194 0.107
11- 1360 0.247 0.109 0.559 0.033 0.192 0.055 0.469 0.088
13- 43 0.045 0.111 0.277 0.042 0.071 0.104 0.262 0.105
14- 22 0.022 0.107 0.272 0.045 0.020 0.067 0.182 0.118
  1. Method numbers are from Table 1.
  2. FNdataset = Feature Number count from dataset
  3. R = Pearson correlation coefficient, values with a are not able to reject the HO: R = 0.
  4. MSE = mean squared error