Table 10 t-test of the difference in prediction error among machine learning methods for high and low reviewer helpfulness groups.
Box office revenue | Number of prediction error data used to compare | Random forests | Decision trees using boosting | k-nearest neighbor method | Discriminant analysis |
|---|---|---|---|---|---|
Week 1 | |||||
Low helpfulness (578)a | 29 | 0.150 | 0.184 | 0.207 | 0.246 |
High helpfulness (578) | 29 | 0.055 | 0.062 | 0.081 | 0.104 |
t-value (p-value) | 4.857 (0.000) | 5.175 (0.000) | 5.205 (0.000) | 4.923 (0.000) | |
Week 2 | |||||
Low helpfulness (578) | 29 | 0.226 | 0.273 | 0.267 | 0.291 |
High helpfulness (578) | 29 | 0.104 | 0.130 | 0.133 | 0.123 |
t-value (p-value) | 4.466 (0.000) | 6.343 (0.000) | 5.774 (0.000) | 5.019 (0.000) | |
Week 3 | |||||
Low helpfulness (578) | 29 | 0.308 | 0.335 | 0.380 | 0.305 |
High helpfulness (578) | 29 | 0.277 | 0.291 | 0.356 | 0.235 |
t-value (p-value) | 1.157 (0.252) | 1.450 (0.153) | 0.696 (0.489) | 2.246 (0.029) | |
