Research Article
Feature Reduction Based on Genetic Algorithm and Hybrid Model for Opinion Mining
Table 17
Confusion matrix of sentiment classification of the two-class reviews using GA algorithm based bagging technique.
| Class/attribute weight | Attribute weight ≥ 0.510 | Attribute weight ≥ 0.400 | Attribute weight ≥ 0.300 | Attribute weight ≥ 0.200 | Attribute weight ≥ 0.100 | Actual neg. | Actual pos. | Actual neg. | Actual pos. | Actual neg. | Actual pos. | Actual neg. | Actual pos. | Actual neg. | Actual pos. |
| Book | Predicted neg. (type II error) | 43 |
08 (8.79) | 61 | 05 (5.49) | 68 | 09 (9.89) | 78 | 12 (13.18) | 82 | 09 (9.89) | Predicted pos. (type I error) | 48 (48.00) | 92 | 30 (30.00) | 95 | 23 (23.00) | 91 | 13 (13.00) | 88 | 09 (9.00) | 91 | Overall error rate (%) | 29.31 | 29.31 | 16.75 | 13.08 | 09.42 | Average accuracy (%) | 70.76 | 70.76 | 83.29 | 86.92 | 90.63 |
| DVD | Predicted neg. (type II error) | 71 | 09 (9.00) | 70 | 07 (7.00) | 72 | 07 (7.00) | 80 | 13 (13.00) | 87 | 11 (11.00) | Predicted pos. (type I error) | 29 (29.29) | 90 | 30 (30.30) | 92 | 28 (28.28) | 92 | 20 (20.20) | 86 | 13 (13.13) | 88 | Overall error rate (%) | 19.09 | 18.59 | 17.58 | 16.58 | 12.06 | Average accuracy (%) | 80.95 | 81.42 | 82.45 | 83.47 | 88.00 |
| Electronics | Predicted neg. (type II error) | 48 | 05 (5.00) | 68 | 05 (5.00) | 73 | 02 (2.00) | 87 | 08 (8.00) | 90 | 07 (7.00) | Predicted pos. (type I error) | 52 (52.00) | 95 | 32 (32.00) | 95 | 27 (27.00) | 98 | 13 (13.00) | 92 | 10 (10.00) | 93 | Overall error rate (%) | 28.50 | 18.50 | 14.00 | 10.50 | 08.50 | Average accuracy (%) | 71.50 | 81.50 | 85.50 | 89.50 | 91.05 |
| Kitchen | Predicted neg. (type II error) | 79 | 45 (49.45) | 77 | 31 (34.06) | 81 | 25 (27.47) | 80 | 21 (23.07) | 72 | 10 (10.98) | Predicted pos. (type I error) | 12 (12.12) | 54 | 14 (14.14) | 68 | 10 (10.10) | 74 | 11 (11.11) | 78 | 19 (19.19) | 89 | Overall error rate (%) | 30.00 | 23.68 | 18.42 | 16.84 | 15.26 | Average accuracy (%) | 70.00 | 76.32 | 81.58 | 83.16 | 84.74 |
| Movie | Predicted neg. (type II error) | 62 | 20 (20.00) | 76 | 16 (16.00) | 91 | 9 (9.00) | 94 | 4 (4.00) | 97 | 3 (3.00) | Predicted pos. (type I error) | 38 (38.00) | 80 | 24 (24.00) | 84 | 9 (9.00) | 91 | 6 (6.00) | 96 | 3 (3.00) | 97 | Overall error rate (%) | 29.00 | 20.00 | 07.00 | 05.00 | 03.00 | Average accuracy (%) | 71.00 | 80.00 | 91.00 | 95.00 | 97.00 |
|
|