This repository creates a XGBoost classification model for COVID-19 using Brazil's OpenDataSUS data as base. The is described in this medium post.
The code is in the two notebooks in the repo, the data can be downloaded from running the GET_DATA.sh on a bash terminal with curl installed. If running on windows or if you don't have curl installed, either install it or download the data from these links:
- https://s3-sa-east-1.amazonaws.com/ckan.saude.gov.br/SRAG/2020/INFLUD-29-07-2020.csv
- save it as data/2020.csv, inside the cloned repo
- https://opendatasus.saude.gov.br/dataset/bd-srag-2012-a-2018
- this contains the other years. Go to 'Explorar' -> 'Baixar' and download each year's data with the name data/{year}.csv
The data can also be downloaded by running make get-data.
The project uses python3 and the libs jupyter, pandas, numpy, plotly, sagemaker, boto3, chart_studio and matplotlib. To install everything, either install it via the requirements.txt file or run make deps.
If running everything at once, run make init.