Table 2 Performance variation of conventional ELM based SLFN with different input size and hidden neurons on volatility forecasting
From: Extreme learning with chemical reaction optimization for stock volatility prediction
Experiment No. | Input Size (n) | Hidden Neuron (Nh) | MAPE (Training) | MAPE (Testing) | Training Time (Sec) |
|---|---|---|---|---|---|
1 | 8 | 15 | 0.073825 | 0.084058 | 46.6312 |
2 | 12 | 18 | 0.050002 | 0.055072 | 57.0700 |
3 | 7 | 14 | 0.049288 | 0.060956 | 46.5883 |
4 | 8 | 22 | 0.100055 | 0.185072 | 68.2664 |
5 | 10 | 19 | 0.058226 | 0.063768 | 47.2374 |
6 | 6 | 16 | 0.032006 | 0.036507 | 58.1268 |
7 | 15 | 12 | 0.050651 | 0.050723 | 46.6685 |
8 | 20 | 15 | 0.700843 | 0.079420 | 56.6251 |
9 | 10 | 8 | 0.615582 | 0.640058 | 36.5455 |
10 | 6 | 25 | 0.079388 | 0.080289 | 49.7146 |