Table 6 The improvement rate of the stacking integrated model compared to the three error indicators of different base learners.
From: Analysis and prediction of infectious diseases based on spatial visualization and machine learning
Algorithm | Evaluation indicators | Stacking vs. Base-learner | |||
|---|---|---|---|---|---|
ARIMA | ELM | SVR | Wavelet | ||
Model I | IMAE(%) | 72.75 | 61.62 | 29.12 | 40.32 |
IRMSE(%) | 74.12 | 74.12 | 25.34 | 46.88 | |
IMAPE(%) | 13.78 | 13.78 | 17.99 | 21.11 | |
Model II | IMAE(%) | 73.34 | 62.45 | 30.65 | 41.61 |
IRMSE(%) | 74.36 | 74.36 | 26.01 | 47.36 | |
IMAPE(%) | 18.91 | 18.91 | 22.87 | 25.81 | |
Model III | IMAE(%) | 80.12 | 71.99 | 48.28 | 56.45 |
IRMSE(%) | 81.03 | 81.03 | 45.27 | 61.06 | |
IMAPE(%) | 34.62 | 34.62 | 37.80 | 40.18 | |
Model IV | IMAE(%) | 78.94 | 70.33 | 45.21 | 53.87 |
IRMSE(%) | 79.04 | 79.04 | 39.53 | 56.97 | |
IMAPE(%) | 32.05 | 32.05 | 35.37 | 37.83 | |
Model V | IMAE(%) | 74.67 | 64.32 | 34.10 | 44.52 |
IRMSE(%) | 74.94 | 74.94 | 27.70 | 48.56 | |
IMAPE(%) | 18.91 | 18.91 | 22.87 | 25.81 | |
Algorithm | Evaluation indicators | Stacking vs. Base-learner | |||
RNN | SGDM-LSTM | Adam-LSTM | RMSProp-LSTM | ||
Model I | IMAE(%) | 59.78 | 52.93 | 53.75 | 43.43 |
IRMSE(%) | 61.23 | 45.02 | 49.89 | 45.70 | |
IMAPE(%) | 20.88 | 16.20 | 15.41 | 8.81 | |
Model II | IMAE(%) | 60.65 | 53.94 | 54.75 | 44.65 |
IRMSE(%) | 61.58 | 45.52 | 50.34 | 46.19 | |
IMAPE(%) | 25.59 | 21.18 | 20.44 | 14.24 | |
Model III | IMAE(%) | 70.65 | 65.65 | 66.25 | 58.72 |
IRMSE(%) | 71.58 | 59.70 | 63.27 | 60.20 | |
IMAPE(%) | 40.00 | 36.45 | 35.85 | 30.85 | |
Model IV | IMAE(%) | 68.91 | 63.61 | 64.25 | 56.27 |
IRMSE(%) | 68.60 | 55.47 | 59.41 | 56.02 | |
IMAPE(%) | 37.65 | 33.96 | 33.33 | 28.14 | |
Model V | IMAE(%) | 62.61 | 56.23 | 57.00 | 47.40 |
IRMSE(%) | 62.46 | 46.77 | 51.47 | 47.42 | |
IMAPE(%) | 25.59 | 21.18 | 20.44 | 14.24 | |
