/DeepOutbreak

Primary LanguagePythonMIT LicenseMIT

DeepOutbreak

Requirements

Use the package manager conda to install required Python dependencies. Note: We used Python 3.7.

conda env create -f requirements.yml

Training ILI model

Enter to ILI folder. To train the model, run this command:

python ./TrainPredict.py

You can set up your own hyperparameter values and data/modules to remove. See help on TrainPredict.py for more details.

e.g.

python ./TrainPredict.py --data linelist --recon_weight 0.005 --alpha 0.3 --_beta 0.01

To evaluate the results, go to evaluate.py and change line 77 for the name of results file (saved in folder rmse_results). Then, run.

python ./evaluate.py

Training COVID model

Enter to COVID folder. To train the model, run this command:

python ./src/training/testbed.py --infile1 ./scripts/specs/base.json --infile2 ./scripts/specs/m1_weekly.json --target death --runs 20 --data_ew 202046 --pred_ew 202046

If you want to get predictions for previous weeks, say epidemic week 40, set prediction week (pred_ew):

e.g.

python ./src/training/testbed.py --infile1 ./scripts/specs/base.json --infile2 ./scripts/specs/m1_weekly.json --target death --runs 20 --data_ew 202046 --pred_ew 202040

To parse and evaluate the results (change line 101 to update prediction week):

python ./src/training/parse_results.py