Disclaimer
This application is fetching data from the sources described below (namely italian dpc for italian data, and Hopkins University for world data). Only a very simplistic and objective analysis is performed: see the explanations given below before giving any significance to the data plotted. You are encouraged to report any problem by opening an issue on github.
- Data source:
- data collected from here
- Data set:
- Series:
- Modifier:
- data collected from here
- Data set:
- Series:
- Modifier:
- data collected from here
- Data set:
- Series:
- data collected from here
- Data set:
- Series:
- data has been extracted from here aura vision lockdown tracker
- Series:
- Smoothing: filter days:
- Press button:
Explanation
A data series is loaded from the selected data set of the given data source. For DPC series a modifier option is available; incremento: daily increment of given series, tasso: daily percentual increment, tasso incremento: daily percentual increment of increment, ratio: percentual ratio between two series or with population. Each series is possibly filtered with the specified smoothing filter; flat: is a running average over the specified number of days in the past, flat centered: as flat but the average is centered on the current day (no delay, but there is an effect on boundary points), log flat: as flat but the logariths of the values are averaged, log flat centered: as flat centered but the logarithms are averaged, binomial: is a filter with binomial coefficients (approximating a gaussian filter). If the series is a ratio of two dataset (or a rate of growth) the filters are applied to each series before computing the ratio. From each series y=y(t) we consider as many data points in the past as specified by the user up to the date specified (or today if no date is specified in the up to date input field). We compute the linear regression of the logarithms of the values: log(y) = mt + q. This gives the exponential fit curve y=exp(mt+q). The average daily increase is the average daily percentual increase: exp(m)-1. Rt is computed assuming 7 days average time of infection exp(7m). The origin is the time when the exponential fit has a unit value: t=-q/m. The growth rate plot shows the percentual increase in each day with respect to the previous day. The growth rate smooth smooth also adds a smoothing filter. The time alignment selection translates each curve in time with respect to the first curve plotted. If prediction shift is selected, each curve is shifted by the number of days required to reach the level of the first curve following the current exponential trend. If history shift is selected, each curve is shifted by the number of days in the past when the first curve had the same value (when the first curve is larger than the current) or in the future when it has the same value of the first curve (when the first curve is smaller than the current). If you press the button set url for this chart the url of the page is changed so that you can save or share a link to the current chart. To perform your own analysis you can download the data in CSV format.