This is the pytorch implementation for class project of AIGS703I(인공지능특론:그래프분석을 위한 기계학습)
This problem is challenging due to the complicated and dynamic spatio-temporal dependencies between different regions in the road network.
Models are mainly based on GraphWaveNet and refer to other state-of-date models.
Dataset: Metr-LA, PeMS-Bay
- Role analysis of traffic graph data
- Community analysis of traffic graph data(Louvain.ipynb)
python train.py --data data/METR-LA --savehorizon True --sheetname Model2_METR-LA_60min --adjdata data/sensor_graph/adj_mx.pkl --batch_size 32 --adjtype doubletransition
python train.py --data data/PEMS-BAY --savehorizon True --sheetname Model2_PEMS-BAY_60min --num_nodes 325 --adjdata data/sensor_graph/adj_mx_bay.pkl batch_size 32 --nhid 16 --adjtype doubletransition
python train.py --data data/METR-LA --savehorizon True --sheetname Model1_METR-LA_60min --adjdata data/sensor_graph/adj_mx.pkl --nhid 16 --batch_size 32 --adjtype transition
python train.py --data data/PEMS-BAY --savehorizon True --sheetname Model1_PEMS-BAY_60min --num_nodes 325 --adjdata data/sensor_graph/adj_mx_bay.pkl --nhid 16 --batch_size 32 --adjtype transition
--batch_size 32
--batch_size 32