FedMetro: Efficient Metro Passenger Flow Prediction via Federated Graph Learning

Official PyTorch implementation of "FedMetro: Efficient Metro Passenger Flow Prediction via Federated Graph Learning"

Data

We collated and provided three real metro data sets from Beijing (BJMetro), Shanghai (SHMetro) and Hangzhou (HZMetro).

Environment

Our experiments are implemented in PyTorch 1.13.1 on Intel(R) Xeon(R) Gold 6230R CPU @ 2.10GHz and four NVIDIA A100 GPUs with 40GB memory.

conda create -n fedmetro "python=3.11"
conda activate fedmetro
bash install.sh

Run

bash run.sh