Influence of the dark figure
In the corona pandemic, many virologists assume that the number of infected people is 100 times greater than the number of people tested positive. If you multiplicate all measured values with this factor (dark figure), this has a considerable influence on the curve fitting, since it brings you closer to the saturation limit.
Without dark figure: | |
Data from: "JHU_DE_Mrz-Apr.csv" Saturation limit: 56 Mio Dark figure: 1 f(t) = 4,559E10/(814,1 + 5,51E7*e^(-0,112*t)) Inflection point W(99,4/28 Mio) Maximum growth rate f'(xw) = 1,5688 Mio 60 Values Coeff.of determin. = 0,82574762 Correlation coeff. = 0,90870656 Standard deviation = 0,90673232 |
![]() |
With dark figure = 100: | |
Data from: "JHU_DE_Mrz-Apr.csv" Saturation limit: 56 Mio Dark figure: 100 f(t) = 4,250E12/(75885 + 5,59E7*e^(-0,119*t)) Inflection point W(55,437/28 Mio) Maximum growth rate f'(xw) = 1,6674 Mio 60 Values Coeff.of determin. = 0,8471621 Correlation coeff. = 0,92041409 Standard deviation = 0,89105973 |
![]() |
With dark figure = 330: | |
Data from: "JHU_DE_Mrz-Apr.csv" Saturation limit: 56 Mio Dark figure: 330 f(t) = 7,82E12/(1,40E5 + 5,59E7*e^(-0,162*t)) Inflection point W(36,931/28 Mio) Maximum growth rate f'(xw) = 2,2714 Mio 60 Values Coeff.of determin. = 0,95628418 Correlation coeff. = 0,97789784 Standard deviation = 0,61100523 |
![]() |
The coefficient of determination is obviously highest in the third example and the curve also seems to match the measured values best. However, one always has to ask where the data modeling ends and the data manipulation begins.