# TrafficFlowPrediction

**Repository Path**: chanzany/TrafficFlowPrediction

## Basic Information

- **Project Name**: TrafficFlowPrediction
- **Description**: Traffic Flow Prediction with Neural Networks(SAEs、LSTM、GRU).
- **Primary Language**: Unknown
- **License**: MIT
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 2
- **Forks**: 0
- **Created**: 2020-12-24
- **Last Updated**: 2022-07-28

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# Traffic Flow Prediction
Traffic Flow Prediction with Neural Networks(SAEs、LSTM、GRU).

## Requirement
- Python 3.6    
- Tensorflow-gpu 1.5.0  
- Keras 2.1.3
- scikit-learn 0.19

pip install Keras==2.1.3
pip install tensorflow==1.5.0

## Train the model

**Run command below to train the model:**

```
python train.py --model model_name
```

You can choose "lstm", "gru" or "saes" as arguments. The ```.h5``` weight file was saved at model folder.


## Experiment

Data are obtained from the Caltrans Performance Measurement System (PeMS). Data are collected in real-time from individual detectors spanning the freeway system across all major metropolitan areas of the State of California.
	
	device: Tesla K80
	dataset: PeMS 5min-interval traffic flow data
	optimizer: RMSprop(lr=0.001, rho=0.9, epsilon=1e-06)
	batch_szie: 256 


**Run command below to run the program:**

```
python main.py
```

These are the details for the traffic flow prediction experiment.

LSTM
explained_variance_score:0.932454
mape:18.851012%
mae:7.858430
mse:109.724229
rmse:10.474933
r2:0.932454
GRU
explained_variance_score:0.939164
mape:26.271195%
mae:7.582251
mse:100.641479
rmse:10.032023
r2:0.938045
SAEs
explained_variance_score:0.927053
mape:23.476228%
mae:11.858915
mse:234.535436
rmse:15.314550
r2:0.855621


| Metrics | MAE | MSE | RMSE | MAPE |  R2  | Explained variance score |
| ------- |:---:| :--:| :--: | :--: | :--: | :----------------------: |
| LSTM | 7.21 | 98.05 | 9.90 | 16.56% | 0.9396 | 0.9419 |
| GRU | 7.20 | 99.32 | 9.97| 16.78% | 0.9389 | 0.9389|
| SAEs | 7.06 | 92.08 | 9.60 | 17.80% | 0.9433 | 0.9442 |

![evaluate](/images/eva.png)

## Reference

	@article{SAEs,  
	  title={Traffic Flow Prediction With Big Data: A Deep Learning Approach},  
	  author={Y Lv, Y Duan, W Kang, Z Li, FY Wang},
	  journal={IEEE Transactions on Intelligent Transportation Systems, 2015, 16(2):865-873},
	  year={2015}
	}
	
	@article{RNN,  
	  title={Using LSTM and GRU neural network methods for traffic flow prediction},  
	  author={R Fu, Z Zhang, L Li},
	  journal={Chinese Association of Automation, 2017:324-328},
	  year={2017}
	}


## Copyright
See [LICENSE](LICENSE) for details.