This project is the code of AAAI 2018 paper When Will You Arrive? Estimating Travel Time Based on Deep Neural Networks. We provide the complete version of code and part of sample data in Beijing. You can replace the sample data with your own data easily. See the samples in data/ for more details. We further provide a pre-trained model in saved_weights/ folder.

Usage:

Model Training

python train.py

Parameters:

• model: The model to train (e.g., AttrTTE, DeepTTE, see models folder)
• task: train/test
• batch_size: the batch_size to train, default 400
• epochs: the epoch to train, default 100
• kernel_size: the kernel size of Geo-Conv, only used when the model contains the Geo-conv part
• pooling_method: attention/mean
• alpha: the weight of combination in multi-task learning
• driver_off: if the driver_off = 1, then all the driver ID is reset to 0, this option is used to show the effectiveness of the driverID embedding.
• week_off: similar with driver_off, used for the weekID embedding
• road_off: whether to use the road information
• log_file: the path of log file

The training log will be recorded to log_file

Model Evaluation

Parameters:

• weight_file: the path of model weight
• result_file: the path to save the result

Example:

Train:
python main.py --model DeepTTE --batch_size 400 --epochs 100 --log_file deeptte_log --pooling_method attention --kernel_size 3 --alpha 0.3

Test:
python main.py --task test --model DeepTTE --batch_size 10 --weight_file ./saved_weights/model_weight --result_file ./result/deeptte.res --pooling_method attention --kernel_size 3