Table 4.
Comparison of software packages based on classification by clustering method. The accuracies of the machine learning models used for the heterogeneous datasets are presented.
| Model | RMS CA (%) | WDMS CA (%) | ODMS CA (%) |
|---|---|---|---|
| Naive Bayes | 60.4 | 67.0 | 80.7 |
| Generalised Linear Model | 60.7 | NA | NA |
| Fast Large Margin | 62.2 | NA | NA |
| Deep Learning/Neural Network | 59.2 | NA | 98.9 |
| Decision Tree | 54.3 | 64.0 | 99.5 |
| Decision table | NA | 69.0 | NA |
| Random Forest | 59.2 | 70.0 | 89.9 |
| Stochastic Gradient Descent | 60.1 | NA | 99.3 |
| Support Vector Machine | 61.3 | 48.0 | 98.4 |
| K-Nearest Neighbours | NA | NA | 99.1 |
| CN2 Induction | NA | NA | 99.5 |
| J48 | NA | 70.0 | NA |
| Average | 59.7 | 64.7 | 95.7 |
Legend: ODMS = Orange Data Mining Software, WDMS = Weka Data Mining Software and RMS = RapidMiner Studio (RMS), NA = Not available, CA = Classification Accuracy.