Table 2. Evaluation of predicted performance according to balanced accuracy.
The balanced accuracies were calculated by 10-fold cross-validation. Values represent the mean ± 10-fold variance. Figures written in bold represent a higher level of balanced accuracy than those of the other classifiers in each class. Figures given in parentheses represent the number of features used in classifiers.
| Kinship ≥ 0.00 | ||||||
| Approach 1 | Approach 2 | |||||
| Class | LogitBoost (83) | KNN (IBk) (80) | SVM (SMO) (86) | LogitBoost (81) | KNN (IBk) (90) | SVM (SMO) (87) |
| D1 | 0.387 ± 0.124 | 0.050 ± 0.025 | 0.000 ± 0.000 | 0.446 ± 0.146 | 0.133 ± 0.104 | 0.017 ± 0.003 |
| D2 | 0.683 ± 0.057 | 0.729 ± 0.039 | 0.810 ± 0.035 | 0.829 ± 0.026 | 0.808 ± 0.044 | 0.742 ± 0.030 |
| D10 | 0.422 ± 0.068 | 0.421 ± 0.088 | 0.701 ± 0.039 | 0.513 ± 0.061 | 0.389 ± 0.114 | 0.612 ± 0.032 |
| D11 | 0.713 ± 0.033 | 0.676 ± 0.107 | 0.695 ± 0.042 | 0.735 ± 0.036 | 0.624 ± 0.100 | 0.625 ± 0.108 |
| D13 | 0.751 ± 0.039 | 0.823 ± 0.024 | 0.905 ± 0.011 | 0.703 ± 0.025 | 0.810 ± 0.027 | 0.913 ± 0.018 |
| D18 | 0.418 ± 0.041 | 0.245 ± 0.071 | 0.277 ± 0.117 | 0.547 ± 0.048 | 0.254 ± 0.032 | 0.291 ± 0.109 |
| D27 | 0.540 ± 0.067 | 0.532 ± 0.065 | 0.513 ± 0.076 | 0.585 ± 0.083 | 0.416 ± 0.051 | 0.521 ± 0.062 |
| D38 | 0.774 ± 0.045 | 0.712 ± 0.074 | 0.682 ± 0.064 | 0.857 ± 0.057 | 0.685 ± 0.051 | 0.697 ± 0.074 |
| D59 | 0.797 ± 0.060 | 0.642 ± 0.069 | 0.648 ± 0.098 | 0.768 ± 0.050 | 0.698 ± 0.047 | 0.677 ± 0.063 |
| D60 | 0.755 ± 0.091 | 0.067 ± 0.044 | 0.145 ± 0.038 | 0.611 ± 0.114 | 0.108 ± 0.034 | 0.361 ± 0.118 |
| D61 | 0.605 ± 0.150 | 0.462 ± 0.160 | 0.150 ± 0.114 | 0.185 ± 0.082 | 0.273 ± 0.044 | 0.000 ± 0.000 |
| D62 | 0.923 ± 0.017 | 0.469 ± 0.064 | 0.475 ± 0.131 | 0.840 ± 0.040 | 0.608 ± 0.062 | 0.486 ± 0.066 |
| D66 | 0.655 ± 0.082 | 0.448 ± 0.170 | 0.340 ± 0.062 | 0.463 ± 0.115 | 0.407 ± 0.038 | 0.411 ± 0.145 |
| D89 | 0.000 ± 0.000 | 0.000 ± 0.000 | 0.000 ± 0.000 | 0.000 ± 0.000 | 0.000 ± 0.000 | 0.000 ± 0.000 |
| D90 | 0.693 ± 0.080 | 0.827 ± 0.038 | 0.527 ± 0.132 | 0.708 ± 0.104 | 0.717 ± 0.068 | 0.689 ± 0.133 |
| D100 | 0.638 ± 0.082 | 0.889 ± 0.035 | 0.663 ± 0.094 | 0.705 ± 0.110 | 0.833 ± 0.125 | 0.746 ± 0.048 |
| D102 | 0.770 ± 0.037 | 0.355 ± 0.055 | 0.683 ± 0.120 | 0.862 ± 0.030 | 0.364 ± 0.038 | 0.702 ± 0.049 |
| D103 | 0.452 ± 0.031 | 0.607 ± 0.095 | 0.492 ± 0.045 | 0.418 ± 0.100 | 0.517 ± 0.081 | 0.468 ± 0.100 |
| D107 | 0.733 ± 0.063 | 0.787 ± 0.045 | 0.757 ± 0.040 | 0.818 ± 0.076 | 0.718 ± 0.080 | 0.795 ± 0.024 |
| D114 | 0.766 ± 0.056 | 0.630 ± 0.077 | 0.678 ± 0.050 | 0.683 ± 0.102 | 0.628 ± 0.055 | 0.747 ± 0.040 |
| Balanced Accuracy | 0.624 ± 0.061 | 0.518 ± 0.067 | 0.507 ± 0.065 | 0.614 ± 0.070 | 0.500 ± 0.060 | 0.525 ± 0.061 |
| Kinship ≥ 0.05 | ||||||
| Approach 1 | Approach 2 | |||||
| Class | LogitBoost (72) | KNN (IBk) (88) | SVM (SMO) (88) | LogitBoost (85) | KNN (IBk) (81) | SVM (SMO) (90) |
| D11 | 0.975 ± 0.006 | 0.940 ± 0.018 | 0.933 ± 0.028 | 0.975 ± 0.006 | 0.980 ± 0.004 | 0.933 ± 0.012 |
| D59 | 0.793 ± 0.103 | 0.904 ± 0.029 | 0.832 ± 0.032 | 0.938 ± 0.010 | 0.848 ± 0.032 | 0.751 ± 0.057 |
| D60 | 0.843 ± 0.029 | 0.150 ± 0.065 | 0.597 ± 0.142 | 0.717 ± 0.073 | 0.204 ± 0.045 | 0.575 ± 0.132 |
| D62 | 0.955 ± 0.009 | 0.767 ± 0.063 | 0.727 ± 0.052 | 0.963 ± 0.006 | 0.693 ± 0.066 | 0.718 ± 0.100 |
| D66 | 0.900 ± 0.024 | 0.661 ± 0.061 | 0.696 ± 0.080 | 0.795 ± 0.116 | 0.591 ± 0.081 | 0.623 ± 0.090 |
| D89 | 0.000 ± 0.000 | 0.000 ± 0.000 | 0.000 ± 0.000 | 0.000 ± 0.000 | 0.000 ± 0.000 | 0.000 ± 0.000 |
| D90 | 0.838 ± 0.035 | 0.802 ± 0.076 | 0.947 ± 0.014 | 0.875 ± 0.045 | 0.806 ± 0.106 | 0.944 ± 0.014 |
| D100 | 0.900 ± 0.044 | 0.941 ± 0.015 | 0.867 ± 0.048 | 0.967 ± 0.011 | 0.889 ± 0.111 | 0.817 ± 0.114 |
| Balanced Accuracy | 0.776 ± 0.031 | 0.646 ± 0.041 | 0.700 ± 0.049 | 0.779 ± 0.033 | 0.626 ± 0.056 | 0.670 ± 0.065 |
| Kinship ≥ 0.10 | ||||||
| Approach 1 | Approach 2 | |||||
| Class | LogitBoost (47) | KNN (IBk) (24) | SVM (SMO) (73) | LogitBoost (64) | KNN (IBk) (28) | SVM (SMO) (50) |
| D59 | 1.000 ± 0.000 | 0.896 ± 0.022 | 0.947 ± 0.007 | 1.000 ± 0.000 | 0.925 ± 0.016 | 0.950 ± 0.013 |
| D62 | 1.000 ± 0.000 | 0.917 ± 0.031 | 0.745 ± 0.051 | 1.000 ± 0.000 | 0.942 ± 0.016 | 0.922 ± 0.017 |
| D66 | 0.942 ± 0.016 | 0.785 ± 0.060 | 0.847 ± 0.046 | 0.930 ± 0.027 | 0.848 ± 0.028 | 0.677 ± 0.108 |
| D89 | 0.000 ± 0.000 | 0.000 ± 0.000 | 0.000 ± 0.000 | 0.000 ± 0.000 | 0.000 ± 0.000 | 0.000 ± 0.000 |
| D100 | 0.950 ± 0.025 | 0.933 ± 0.020 | 0.811 ± 0.050 | 0.963 ± 0.012 | 0.900 ± 0.026 | 0.622 ± 0.129 |
| Balanced Accuracy | 0.778 ± 0.008 | 0.706 ± 0.026 | 0.670 ± 0.031 | 0.779 ± 0.008 | 0.723 ± 0.017 | 0.634 ± 0.053 |
| Kinship ≥ 0.15 | ||||||
| Approach 1 | Approach 2 | |||||
| Class | LogitBoost (72) | KNN (IBk) (88) | SVM (SMO) (88) | LogitBoost (85) | KNN (IBk) (81) | SVM (SMO) (90) |
| D59 | 0.950 ± 0.025 | 0.989 ± 0.001 | 0.988 ± 0.002 | 1.000 ± 0.000 | 1.000 ± 0.000 | 0.975 ± 0.006 |
| D100 | 0.852 ± 0.031 | 0.933 ± 0.020 | 0.858 ± 0.037 | 0.950 ± 0.025 | 0.875 ± 0.051 | 0.775 ± 0.068 |
| Balanced Accuracy | 0.901 ± 0.028 | 0.961 ± 0.010 | 0.923 ± 0.019 | 0.975 ± 0.013 | 0.938 ± 0.026 | 0.875 ± 0.037 |