Table 2. The accuracy of four different tree induction models (each ran with four criteria, Accuracy, Gain Ratio, Gini Index and Info Gain) on 11 datasets [original protein features dataset (FCdb) as well as 10 datasets generated by trimming (filtering) the original FCdb dataset by attribute weighting algorithms) computed by 10-fold cross validation.
| Decision Tree | Decision Tree Parallel | Decision Tree Stump | Decision Tree Random Forest | |||||||||||||
| Dataset Filtered by | Accuracy | Gain Ratio | Gini Index | Info Gain | Accuracy | Gain Ratio | Gini Index | Info Gain | Accuracy | Gain Ratio | Gini Index | Info Gain | Accuracy | Gain Ratio | Gini Index | Info Gain |
| Chi Squared | 99.52 | 99.52 | 99.52 | 99.52 | 94.22 | 99.27 | 99.46 | 99.47 | 71.63 | 48.48 | 71.18 | 71.98 | 88.90 | 91.72 | 99.51 | 99.54 |
| Info Gain | 99.33 | 99.33 | 99.33 | 99.33 | 89.30 | 99.56 | 99.51 | 99.33 | 71.63 | 49.80 | 71.18 | 71.98 | 86.18 | 95.92 | 99.52 | 99.44 |
| Deviation | 70.73 | 70.73 | 70.73 | 70.73 | 55.01 | 70.73 | 96.78 | 96.80 | 50.65 | 50.73 | 58.85 | 58.85 | 55.13 | 70.96 | 96.65 | 96.80 |
| Gini Index | 99.24 | 99.24 | 99.24 | 99.24 | 89.19 | 99.39 | 99.29 | 99.32 | 71.63 | 49.80 | 71.18 | 71.98 | 82.47 | 94.45 | 99.43 | 99.14 |
| Info Gain Ratio | 99.43 | 99.43 | 99.43 | 99.43 | 91.69 | 99.40 | 99.35 | 99.25 | 71.63 | 48.48 | 71.18 | 71.98 | 87.37 | 93.12 | 99.01 | 99.33 |
| PCA | 70.73 | 70.73 | 70.73 | 70.73 | 55.01 | 70.73 | 96.78 | 96.80 | 50.65 | 50.73 | 58.85 | 58.85 | 55.13 | 70.96 | 96.65 | 96.80 |
| Relief | 99.35 | 99.35 | 99.35 | 99.35 | 87.58 | 99.31 | 99.37 | 99.22 | 71.63 | 49.80 | 71.18 | 71.98 | 87.88 | 96.64 | 99.24 | 98.87 |
| Rule | 99.47 | 99.47 | 99.47 | 99.47 | 91.36 | 99.37 | 99.24 | 99.39 | 71.63 | 49.80 | 71.18 | 71.86 | 82.51 | 93.81 | 99.36 | 98.86 |
| Uncertainty | 99.40 | 99.40 | 99.40 | 99.40 | 91.70 | 99.40 | 99.32 | 99.25 | 71.63 | 49.80 | 71.18 | 71.98 | 88.62 | 95.01 | 99.70 | 99.65 |
| SVM | 98.99 | 98.99 | 98.99 | 98.99 | 93.64 | 92.48 | 99.09 | 99.09 | 70.99 | 46.40 | 70.05 | 72.03 | 86.13 | 89.00 | 98.56 | 98.06 |
| FCdb (original protein features dataset) | 99.06 | 96.87 | 74.36 | 46.04 | 76.13 | 96.87 | 74.93 | 74.50 | 48.48 | 90.00 | 58.38 | 58.38 | 97.65 | 99.31 | 98.34 | 97.73 |
This table presents the accuracy percentage of Tree Induction models (Decision Tree, Decision Tree Parallel, Decision Stump, Random Forest and Random Tree) run with four different criteria (Gain Ratio, Information Gain, Gini Index and Accuracy). The lowest and highest accuracies have been highlighted.