Table 6 Prediction accuracy of the training and test data sets using Random Forest and GEMS-SVM
| Â | Compounds | Sample size | Prediction Accuracy | |
|---|---|---|---|---|
| Â | Â | Â | Random Forest | GEMS |
Training | 1,2-dichlorobenzene | 34 | 91.20% | 85.30% |
| Â | 1,4-dichlorobenzene | 36 | 94.40% | 94.40% |
| Â | bromobenzene | 36 | 100% | 94.40% |
| Â | diquat | 72 | 86.10% | 88.90% |
| Â | galactosamine | 36 | 83.30% | 86.10% |
| Â | monocrotaline | 32 | 84.40% | 84.40% |
| Â | N-nitrosomorpholine | 36 | 94.40% | 100% |
| Â | thioacetamide | 36 | 86.10% | 83.30% |
| Â | All training | 318 | 89.60% | 89.60% |
Test | acetaminophen (NCT) | 36 | 88.90% | 88.90% |
| Â | acetaminophen (NTP) | 64 | 87.50% | 87.50% |
| Â | acetaminophen (Informatics Challenge) | 108 | 90.70% | 89.80% |
| Â | carbon tetrachloride | 72 | 77.80% | 77.80% |
| Â | Ally Alcohol | 95 | 64.20% | 59% |