Table 8.
Sentiment accuracy differences on polynomial dataset
| Algorithm | Variation between train–test split percentages | |||||
|---|---|---|---|---|---|---|
| 10–20% | 20–30% | 30–40% | 40–50% | 50–60% | 60–70% | |
| KNN | + 2.53 | + 1.91 | + 1.33 | + 0.93 | + 0.37 | + 0.86 |
| DT | + 0.86 | − 0.05 | − 0.17 | + 0.75 | − 0.22 | + 0.48 |
| SVM | + 3.56 | + 2.27 | + 1.53 | + 1.60 | + 0.31 | + 0.97 |
| NB | + 3.13 | + 0.63 | + 1.00 | + 1.79 | − 0.61 | + 0.81 |
| Voting | + 2.71 | + 1.79 | + 1.62 | + 1.33 | − 0.12 | + 0.95 |
| Bagging (KNN) | + 2.69 | + 1.46 | + 1.65 | + 1.15 | + 0.30 | + 0.75 |
| Bagging (DT) | + 0.60 | − 0.40 | + 0.66 | + 0.25 | − 0.22 | + 0.62 |
| Bagging (SVM) | + 3.56 | + 2.28 | + 1.42 | + 1.74 | + 0.24 | + 1.01 |
| Bagging (NB) | + 3.01 | + 0.82 | + 0.97 | + 1.77 | -0.66 | + 0.95 |