Table 2 Model architecture.
From: Demand forecasting of smart tourism integrating spatial metrology and deep learning
Structure | Description |
|---|---|
Input layer | Include multiple time series input variables and space factors |
Multilayer LSTM encoder | Two-layer stacked LSTM unit, the output of each layer is sent to the lower layer or output layer, which improves the depth of model fitting |
Dropout layer | Used to prevent over-fitting and improve the generalization ability of the model |
Fully connected layer | Map the time sequence code output by LSTM to the target predicted value |
Output layer | Output the prediction value of tourism demand in the future time |
