Skip to main content

Table 4 Classification accuracy of the support vector machine with three different kernels with default parameters

From: PredCRG: A computational method for recognition of plant circadian genes by employing support vector machine with Laplace kernel

Dataset Kernel Sensitivity Specificity Accuracy Precision
Q1 Linear 59.24 \(\pm\) 0.90 63.02 \(\pm\) 3.42 61.13 \(\pm\) 1.66 61.63 \(\pm\) 2.16
  Laplace 58.86 \(\pm\) 1.71 63.21 \(\pm\) 2.42 61.04 \(\pm\) 1.86 61.56 \(\pm\) 2.07
  Radial 68.64 \(\pm\) 1.55 51.86 \(\pm\) 3.42 60.25 \(\pm\) 1.54 58.81 \(\pm\) 1.55
Q2 Linear 63.32 \(\pm\) 2.06 66.21 \(\pm\) 1.39 64.76 \(\pm\) 1.48 65.19 \(\pm\) 1.43
  Laplace 64.07 \(\pm\) 1.83 64.32 \(\pm\) 2.11 64.20 \(\pm\) 1.49 64.25 \(\pm\) 1.57
  Radial 67.52 \(\pm\) 3.67 60.62 \(\pm\) 1.30 64.07 \(\pm\) 1.93 63.14 \(\pm\) 1.47
Q3 Linear 66.10 \(\pm\) 4.34 63.89 \(\pm\) 2.85 65.01 \(\pm\) 2.20 64.67 \(\pm\) 1.94
  Laplace 70.56 \(\pm\) 4.25 60.81 \(\pm\) 2.61 65.69 \(\pm\) 1.93 64.29 \(\pm\) 1.56
  Radial 67.61 \(\pm\) 3.52 60.75 \(\pm\) 4.74 64.18 \(\pm\) 1.11 63.36 \(\pm\) 1.78
Q4 Linear 59.26 \(\pm\) 2.29 57.94 \(\pm\) 3.71 58.61 \(\pm\) 2.50 58.53 \(\pm\) 2.63
  Laplace 64.91 \(\pm\) 2.68 53.11 \(\pm\) 2.27 59.01 \(\pm\) 1.71 58.06 \(\pm\) 1.48
  Radial 59.70 \(\pm\) 2.52 60.29 \(\pm\) 1.98 60.01 \(\pm\) 1.51 60.05 \(\pm\) 1.45
  1. Classification was made with each sub dataset and performance metrics were computed following repeated cross validation where the experiment was repeated 100 times. In terms of accuracy, performances are higher for the Laplace kernel for Q2 and Q3 sub-datasets, whereas linear and RBF kernel performed better in Q1 and Q4 respectively. Performance metrics are higher for Q2 and Q3 sub-datasets than that of Q1 and Q4. The accuracies are seen to be more stable for RBF kernel, barring few exceptions