COVID-EENet

COVID-EENet: Predicting Fine-Grained Impact of COVID-19 on Local Economies (AAAI-22)

About

• Source code and datasets of the paper COVID-EENet: Predicting Fine-Grained Impact of COVID-19 on Local Economies.
• Since dataset from BCCard is not open to public, we only provide epidemic-view feature and the physical distance dataset of geography-view feature.

Installation

Requirements

• Ubuntu 16.04.7 LTS
• python 3.8 (Recommend Anaconda)
• Pytorch >= 1.9.0

Usage

• Run python main.py to train COVID-EENet

python main.py -h
usage: main.py [-h] [--model_name MODEL_NAME] [--fname FNAME] [--pred_len PRED_LEN] [--cuda CUDA] [--train] [--test] [--save_prediction] [--save_metric_result]
COVIDEENet
optional arguments:
  -h, --help                 show this help message and exit
  --model_name MODEL_NAME    type one of the comparing algorithms including COVID-EENet
  --fname FNAME              type the file name of the parameters, predictions, experiment results
  --pred_len PRED_LEN        type the predicting length of algorithms
  --cuda CUDA                type the number of gpu
  --train                    type when you train the model
  --test                     type when you validate the model
  --save_prediction          type when you save the predictions of the algorithms
  --save_metric_result       type when you save the experiment results of the algorithms

After training the model, you can find

• the learned parameters in directory models_state_dict
• the predictions in directory model_prediction
• the experiment results in directory RMSE_district_buz_pairs

Hyperparameters

Please check the hyperparameters of COVID-EENet defined in Config.py and supplementary material.