Table 2 Details of hyperparameter tuning for each method. xe+y means x × 10y and xe−y means x × 10−y
Method | Hyper-parameter | Candidates | Used hyperparameters | |||
|---|---|---|---|---|---|---|
Order information (−) | Order information (+) | |||||
Oversampling (−) | Oversampling (+) | Oversampling (−) | Oversampling (+) | |||
LR | C | 1e− 4, 1e −3, 1e −2, 1e −1, 1.0, 1e+1, 1e+2, 1e+3, 1e+4 | 1e+2 | 1e −2 | 1.0 | 1.0 |
| Â | L1 ratio | 0.0, 0.5, 1.0 | 1.0 | 0.5 | 1.0 | 1.0 |
GBDT | Iterations | 500, 1,000, 1,500 | 500 | 1,000 | 1,500 | 1,000 |
LSTM | Learning rate | 5e −6, 1e −5, 2e −5, 3e −5 | 5e −6 | 1e −5 | 5e −6 | 5e −6 |
| Â | Epochs | 5, 10, 15, 20, 25, 30 | 20 | 5 | 20 | 25 |
BERT | Learning rate | 1e −5, 2e −5, 3e −5, 4e −5, 5e −5 | 5e −5 | 5e −5 | 4e −5 | 1e −5 |
| Â | Epochs | 1, 2, 3, 4, 5 | 3 | 1 | 3 | 1 |