diodz/ml-covid

We attempt to model the number of cases of COVID-19 as well as number of deaths from the disease. We combine data on cases from the John Hopkins Coronavirus Resource Center, economic and healthcare system variables from the World Bank, and policy variables from the Oxford’s University Covid-19 Government Response Tracker.

Unsupervised ML to Predict Covid-19 Cases and Deaths

CAPP 30254

The objective is to model the cases of COVID-19 and related deaths at the country level. Data from John Hopkins University on cases and deaths, World Bank on country data, and Oxford University on policy measures are used to train the model.

Breakdown of folders:

data: contains cleaned datasets

data: WB data: contains World Bank source data

predictions: python pickle files of predicted data

scripts: contains scripts that clean and prepare the data, train models, output predictions, visualize predictions

trained_classifiers: contains pickle files of the various trained models

visualizations: contains file output from various scripts

Instructions to run the code:

Please run the jupyter noteboooks in the following order:

1. covid_dataset.ipynb
2. model_fitting.ipynb
3. MLP Classifier.ipynb
4. Model_output.ipynb