Table 4 Prediction error comparison of tourist flow by different models in Jiuzhaigou.
From: Combined CNN-BiLSTM-Att tourism flow prediction based on VMD-MWPE decomposition reconstruction
Model | 3 days | 7 days | 15 days | ||||||
|---|---|---|---|---|---|---|---|---|---|
MAPE (%) | BRMSE (%) | R2 | MAPE (%) | BRMSE (%) | R2 | MAPE (%) | BRMSE (%) | R2 | |
ARIMA | 13.84 | 20.64 | 0.727 | 22.34 | 37.72 | 0.614 | 35.99 | 77.20 | 0.498 |
SVM | 26.16 | 33.64 | 0.523 | 38.74 | 49.02 | 0.428 | 51.81 | 63.64 | 0.335 |
DT | 3.11 | 4.68 | 0.986 | 5.17 | 10.09 | 0.888 | 9.17 | 24.35 | 0.436 |
RF | 4.11 | 6.93 | 0.969 | 4.60 | 8.15 | 0.926 | 7.76 | 18.60 | 0.62 |
DNN | 2.19 | 3.23 | 0.993 | 5.46 | 11.01 | 0.866 | 6.85 | 15.72 | 0.729 |
LSTM | 2.67 | 3.51 | 0.991 | 4.83 | 8.46 | 9.03 | 6.24 | 13.86 | 0.772 |
XGBoost | 2.91 | 5.02 | 0.984 | 4.30 | 9.05 | 0.91 | 7.50 | 17.79 | 0.653 |
BiLSTM | 2.24 | 3.37 | 0.993 | 4.28 | 8.86 | 0.945 | 6.36 | 14.21 | 0.758 |
CNN-LSTM | 2.45 | 3.56 | 0.992 | 4.02 | 8.24 | 0.934 | 6.18 | 13.49 | 0.793 |
CNN-BiLSTM | 2.16 | 3.37 | 0.994 | 3.89 | 7.92 | 0.952 | 5.94 | 11.43 | 0.831 |
CNN-BiLSTM-Att | 1.93 | 2.67 | 0.996 | 3.49 | 7.19 | 0.963 | 5.68 | 10.25 | 0.862 |
