@TenYearNote @anilvohra69 I will state, categorically, that this analysis is 100% wrong. If you accumulate a time-series before regressing on it you will get spuriously high correlations- indeed, asymptotically approaching 1- even if the underlying process is highly variable.
@TenYearNote @anilvohra69 Let's assume that the number of infections per day is given by 100 + t + noise of s.d. 50. Then, the R^2 of the best possible fit should be around 33%, and indeed regressing on the cases of infected per day gets around 35% R^2, as we would expect. 
@TenYearNote @anilvohra69 However, if you accumulate the data and have your dependent variable be the total number of infected cases to date, suddenly your R^2- for a process we know is 67% random noise- jumps to 99%+, for totally spurious reasons.
@TenYearNote @anilvohra69 This is such a basic error to make- I would mark down a undergrad in a first-year stats course for doing it- that anyone seriously claiming to be a statistician who makes it is either manifestly unqualified for their job or lying to you to be duplicitous.
