(04-04-2020, 05:56 PM)M T Wrote: The very noisy ones that had the exceptional 202 daily number on the day just before your set of 5, which distorts things greatly. (If you'd chosen 6 days, the average would have been 83.7 in the most recent set versus 54.2 in the previous set)I deliberately did not choose 6 days because it would have been wrong: the county made clear that the 202 on 3/30 was not real; it included results from previous days that had not been reported when they should have been (they probably did not get any reporting from some sites on 1 or more preceding days). So I averaged the 202 with the 4 preceding days; to average the 202 with any subsequent days would have been inappropriate because the average would have transferred some of the new cases from before 3/30 to after 3/30 where they did not belong. The "distortion" caused by the extreme 3/30 number is avoided by averaging it with the preceding days.
What I meant by arbitrary is that my choice of a 5-day average was not random, it was dictated by that reporting glitch.