XeniaLLL / HierST

The open-source implementation of the paper "HierST: A Unified Hierarchical Spatial-temporal Framework for COVID-19 Trend Forecasting"

HierST

Source code for the paper, "HierST: A Unified Hierarchical Spatial-temporal Framework for COVID-19 Trend Forecasting", accepted by CIKM 2021.

Setup

• Install CUDA 10.1
• Run setup_py.sh to install necessary packages

Usage

cd src/
# pull data
python data_utils.py
# train models
python run_models.py --forecast_date 2021-10-24
# ensemble
python run_ensemble.py --forecast_date 2021-10-24
# check the output file, '../outputs/2021-10-24_forecasts.csv'