PM2.5-GNN: A Domain Knowledge Enhanced Graph Neural Network For PM2.5 Forecasting
- Download dataset KnowAir from Google Drive or Baiduyun with code
t82d.
Python 3.7.3
PyTorch 1.7.0
PyG: https://github.com/rusty1s/pytorch_geometric#pytorch-170
pip install -r requirements.txtopen config.yaml, do the following setups.
- set data path after your server name. Like mine.
filepath:
GPU-Server:
knowair_fp: /data/wangshuo/haze/pm25gnn/KnowAir.npy
results_dir: /data/wangshuo/haze/pm25gnn/results- Uncomment the model you want to run.
# model: MLP
# model: LSTM
# model: GRU
# model: GC_LSTM
# model: nodesFC_GRU
model: PM25_GNN
# model: PM25_GNN_nosub- Choose the sub-datast number in [1,2,3].
dataset_num: 3- Set weather variables you wish to use. Following is the default setting in the paper. You can uncomment specific variables. Variables in dataset KnowAir is defined in
metero_var.
metero_use: ['2m_temperature',
'boundary_layer_height',
'k_index',
'relative_humidity+950',
'surface_pressure',
'total_precipitation',
'u_component_of_wind+950',
'v_component_of_wind+950',]python train.pyPaper: https://dl.acm.org/doi/10.1145/3397536.3422208
@inproceedings{10.1145/3397536.3422208,
author = {Wang, Shuo and Li, Yanran and Zhang, Jiang and Meng, Qingye and Meng, Lingwei and Gao, Fei},
title = {PM2.5-GNN: A Domain Knowledge Enhanced Graph Neural Network For PM2.5 Forecasting},
year = {2020},
isbn = {9781450380195},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
url = {https://doi.org/10.1145/3397536.3422208},
doi = {10.1145/3397536.3422208},
abstract = {When predicting PM2.5 concentrations, it is necessary to consider complex information sources since the concentrations are influenced by various factors within a long period. In this paper, we identify a set of critical domain knowledge for PM2.5 forecasting and develop a novel graph based model, PM2.5-GNN, being capable of capturing long-term dependencies. On a real-world dataset, we validate the effectiveness of the proposed model and examine its abilities of capturing both fine-grained and long-term influences in PM2.5 process. The proposed PM2.5-GNN has also been deployed online to provide free forecasting service.},
booktitle = {Proceedings of the 28th International Conference on Advances in Geographic Information Systems},
pages = {163–166},
numpages = {4},
keywords = {air quality prediction, graph neural network, spatio-temporal prediction},
location = {Seattle, WA, USA},
series = {SIGSPATIAL '20}
}
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent