27 January 2021
Code optimized, charts tweaks, particularely automation of annotation spacing. FB Prophet model tweaked to imporve earlier convergance Reporting period trimmed to start from July 2020 as charts were getting crowded with data from the first wave Any suggestions are welcome!
29 October 2020
Dashboards optimized, and more charts added, particularely age group break down of infections.
13 August: Updated charts, added a counter with 'Days with under 100 infections reported' as this is being an important milestone for the province after Phase III re-opening included Toronto, Peel, and Windsor regions.
9 July: COVID dashboards in Federal facilities, US (NY&VA), Canada (ON) updated. Comments in the US Federal facilities updated to make it easier to understand
18 June:
- Updated charts with latest data (US & Canada)
- Fixed date formatting in the charts
- New Plotly charts
17 May: Added COVID 19 projects in Federal Prisons (Using the Marshal Project Data) & Ontario, CA (GVT of Ontario Data). Expanded furher graphics & use of plotly. For Canadian Data, I decided to use data from the Government of Ontario as being the most reliable.
7 May: FB Prophet forecasting method fine tuned with MCMC samples enabled and a good model for NY was generated. ARIMA forecast still unusable. A reporting issue in NY impacted the data as NYT started to report suspected COVID-19 deaths rather than only the ones confirmed by testing.
3 May 2020: I attempted ARIMA and FB Prophet models to forecast future infections. Both models behaved badly!
Exploratory Data Analysis using open source datasets such as NYT I'm using an open-source dataset available from the New York Time: 'https://github.com/nytimes/covid-19-data/raw/master/us-states.csv' / https://github.com/nytimes/covid-19-data
The aim is to generate simple charts to visualize and help us understand the trend of infection growth / decrease and most improtantly seeing if states are flattening the curve.
I'll try to update and comment the code regularely, nonetheless, it should be clear to the reader. Happy to answer any questions.