This project is a nonrigorous yet convenient attempt to estimate the number of COVID-19 cases in selected countries in 2020. All data used are from Worldometer. Currently I am working on making the model more predictive specifically for number of coronavirus cases.

Nobody^* in the West saw COVID-19 coming for them when the outbreak started in China. At the inception of this project (end of March), many developing countries, including India and many in Africa, may be similarly unaware of the impending thread. It is troubling to think how much these places, without advanced medical capabilities like the US or China, may suffer when the outbreak hits. Through this project, I hope to contribute to public awareness of the concern and efforts to mitigate it.

Pull the repo with

git clone https://github.com/Yijia-Chen/covid19-trends.git

Then

cd covid19-trends

To crawl data from the Worldometer site, run

scrapy runspider scraper.py -o data/countries.csv -t csv

You should now see a folder named "data" containing case data of countries selected. Select different countries by changing the list at the top of scraper.py. To then fit models to data crawled, run

python3 model.py <number of days after today you wish to predict>

Remember to replace the text in triangular brackets above with a number and remove brackets. You should now see "plots" and "preds", which store past data and future predictions respectively. The model in model.py is a logistic function; you may play with other functions of your choice and see how well it fits the data.

*: By nobody I mean very few who was heard or believed at the time

If you would like to contribute to this project or just have an cool idea, feel free to contact me at yijia[dot]chen[at]berkeley[dot]edu, and I am more than happy to discuss and we can potentially collaborate.