Use ARIMA, RNN ,LSTM ,Transformer to predict ECG data
ARIMA: ARIMA_local.ipynb
RNN&LSTM single step: LSTM_one_step_local.ipynb
RNN&LSTM multistep step: LSTM_multiple_step_local.ipynb
Transformer: ./informer/informer.ipynb
Figure 3. An predict of the ECG data.
Fitting result of Transformer model:
| Input Length | Hidden unit | Forecast Lenth | Test RMSE | Test R2 |
|---|---|---|---|---|
| 1 | 128 | 1 | 0.004607 | 0.96190 |
| 5 | 128 | 1 | 0.035841 | 0.94453 |
| 5 | 512 | 1 | 0.039515 | 0.93254 |
| 8 | 128 | 1 | 0.070479 | 0.78558 |
| 10 | 128 | 1 | 0.06866 | 0.79652 |
| 10 | 512 | 1 | 0.01472 | 0.61062 |
| 15 | 128 | 1 | 0.021716 | 0.63117 |
| 20 | 128 | 1 | 0.021053 | 0.20606 |
| 25 | 128 | 1 | 0.021716 | 0.15371 |
| 25 | 512 | 1 | 0.019989 | 0.28464 |
| 25 | 512 | 5 | 0.012705 | 0.71109 |
| 30 | 128 | 1 | 0.021278 | 0.18980 |
| 50 | 512 | 25 | 0.009684 | 0.83293 |
| 80 | 512 | 25 | 0.012824 | 0.70580 |
| 250 | 512 | 25 | 0.022158 | 0.12177 |