File size: 1,598 Bytes
0061502
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
---
tags:
- traffic-forecasting
- time-series
- graph-neural-network
- graph-wavenet
datasets:
- metr-la
---

# Graph-WaveNet Model - METR-LA

Graph WaveNet for traffic speed forecasting, combining graph convolution with dilated causal convolution.

## Model Description

This model uses a graph neural network architecture that combines:
- Adaptive adjacency matrix learning
- Spatial graph convolution for capturing spatial dependencies
- Temporal convolution with dilated causal convolutions
- Multi-scale temporal receptive field

## Evaluation Metrics

- **Test MAE (15 min)**: 2.4767
- **Test MAPE (15 min)**: 0.0618
- **Test RMSE (15 min)**: 4.5875


## Dataset

**METR-LA**: Traffic speed data from highway sensors.

## Usage

```python
from utils.gwnet import load_from_hub

# Load model from Hub
model = load_from_hub("METR-LA")

# Get predictions
import numpy as np
x = np.random.randn(10, 12, 207, 2)  # (batch, seq_len, nodes, features)
predictions = model.predict(x)
```

## Training

Model was trained using the Graph-WaveNet implementation with default hyperparameters.

## Citation

If you use this model, please cite the original Graph WaveNet paper:

```bibtex
@inproceedings{wu2019graph,
  title={Graph WaveNet for Deep Spatial-Temporal Graph Modeling},
  author={Wu, Zonghan and Pan, Shirui and Long, Guodong and Jiang, Jing and Zhang, Chengqi},
  booktitle={Proceedings of the 28th International Joint Conference on Artificial Intelligence},
  pages={1907--1913},
  year={2019}
}
```

## License

This model checkpoint is released under the same license as the training code.