Analyis of the evolution of covid-19 cases with time
Analysis of recent coronavirus data for different countries and regions.
Main report is here: https://jmoldon.github.io/coronavirus_analysis/cvid_report.html
Data comes from Johns Hopkins University that is updated daily.
Some sample plots
Response comparison China vs the rest
- Need to change the source database to https://github.com/datasets/covid-19 due to a mismatch with the previous database
- Fix bug related with how to read user-defined initial guess of parameters
- New function to fit. Now it can be pure exponential, logistic or double logistic (two contatenated logistics to account for changes in the propagation rate)
- The new option has more free parameters, so it is more sensitive to bad data (like large jumps) and initial conditions.
- Add plot of deaths per day for 12 countries combined
- Fix date tick labels to be one per week (on Mondays)
- More clear date tick label format
- Add plots for Spain by Comunidades Autónomas
- Added prediction to the number of deaths per day plot
- Changed zorder of Active plots to be lower
- Have to change the source of data because JHU has changed the format unconsistently and have removed the Recovered data
- Add increments to all predictions
- Add stripes to show rate of doubling cases every 2 days
- Fix bug with case differences in the tables
- Added USA per State in the global summary
- Major refactor of some functions
- Now death rate (per day) is plotted
- Mayor refactoring and new functions.
- Now the report displays predicted number of cases for the next 5 days
- Included USA as aggregated country
- Included Poland
- bound_threshold parameter to manually disable confidence intervals
- Add global plots of evolution of most affected countries over time
- Better determination of exponential curve.
- exponential or logistic function is automatically selected based on goodness of fit
- Add confidence interval calculation. Add Germany