Table 4 Accuracy (%) for different feature sets and different classification techniques
Feature set | Base learners | Ensemble learning methods | ||||||
|---|---|---|---|---|---|---|---|---|
C4.5 | NN | SVM | NB | RS-C4.5 | RS-NN | RS-SVM | RS-NB | |
(a) Segmenting type s = like | ||||||||
 F1 | 60.00 | 57.50 | 60.50 | 60.50 | 64.00 | 64.50 | 65.50 | 64.50 |
 F1 + F2 | 57.00 | 54.50 | 58.50 | 53.00 | 68.00 | 68.50 | 68.50 | 59.50 |
 F1 + F2 + F3 | 57.00 | 54.50 | 58.50 | 52.50 | 68.00 | 69.50 | 69.50 | 58.50 |
 F1 + F2 + F3 + F4 | 62.00 | 66.00 | 69.00 | 65.50 | 77.00 | 79.00 | 93.00 | 84.50 |
(b) Segmenting type s = dislike | ||||||||
 F1 | 55.00 | 57.00 | 49.50 | 47.00 | 56.00 | 53.50 | 54.00 | 28.50 |
 F1 + F2 | 58.50 | 59.00 | 57.50 | 24.00 | 64.00 | 67.00 | 65.50 | 44.50 |
 F1 + F2 + F3 | 58.50 | 59.00 | 57.50 | 24.00 | 63.50 | 68.50 | 65.00 | 44.50 |
 F1 + F2 + F3 + F4 | 68.00 | 71.50 | 74.00 | 65.50 | 86.00 | 93.50 | 94.00 | 85.50 |
(c) Segmenting type s = sum | ||||||||
 F1 | 67.50 | 66.00 | 60.50 | 64.50 | 72.50 | 68.00 | 63.00 | 63.50 |
 F1 + F2 | 71.00 | 65.00 | 66.00 | 62.00 | 74.50 | 59.50 | 72.00 | 71.50 |
 F1 + F2 + F3 | 71.00 | 66.50 | 66.00 | 62.00 | 72.00 | 59.00 | 70.50 | 71.50 |
 F1 + F2 + F3 + F4 | 62.00 | 62.00 | 67.00 | 58.50 | 76.50 | 50.00 | 82.50 | 79.50 |
(d) Segmenting type s = portfolio | ||||||||
 F1 | 59.50 | 60.00 | 60.00 | 57.00 | 59.00 | 57.00 | 61.50 | 66.00 |
 F1 + F2 | 65.50 | 62.50 | 63.50 | 60.00 | 71.50 | 65.00 | 77.50 | 74.00 |
 F1 + F2 + F3 | 65.50 | 57.00 | 63.50 | 60.00 | 73.50 | 69.50 | 70.50 | 75.00 |
 F1 + F2 + F3 + F4 | 64.50 | 70.50 | 72.50 | 76.50 | 80.00 | 87.00 | 94.50 | 88.50 |