User talk:Dudzcom

You keep reverting official numbers to Wikipedia chart numbers. These numbers are calculated by Wikipedia, harboring room for bias and miscalculation. There are many reputable case lists with credible sources to back the information. The numbers you are using are obviously strayed from official numbers, and I would recommend reverting your edits. Thanks for understanding, GyozaDumpling (talk) 04:40, 14 March 2020 (UTC)[reply]

@GyozaDumpling:, I posted a reply in https://en.wikipedia.org/wiki/Template_talk:2019%E2%80%9320_coronavirus_pandemic_data/United_States_medical_cases_chart dudzcom (talk) 04:51, 14 March 2020 (UTC)[reply]

@GyozaDumpling:, Also, what makes those numbers official? What is more official than the numbers released by the states themselves? The Johns Hopkins website does not cite sources for their values.

Regarding the total for Georgia at https://en.wikipedia.org/wiki/Template:2019%E2%80%9320_coronavirus_pandemic_data/United_States_medical_cases : I am getting it by summing the column. You are right that the total should be 620, but that means that one or more of the individual days' totals is too high. Which of those day(s) should have a lower figure? — Preceding unsigned comment added by Mark Taylor (talk • contribs) 04:52, 23 March 2020 (UTC)[reply]

Mark Taylor, I understand the issue. I have built many of these types of daily (or hourly/minutely) historical records before, and here's my approach: The value of highest importance is the "summation total", either the state summation or the daily summation; if the summation total does not match records, then it will undermine the trustworthiness of the table - people will notice. We can't realistically go back in time to figure out which day's number is wrong; so the cleanest approach is to choose a day and "eat" the delta - let the daily value be off for a day for that state... just make sure that the totals are correct. Honestly, after a week, no one will care that the value isn't precise, and they will not have any reason to believe that any of the underlying values were imprecise. And as for the daily summation across states, the value we have just "eaten" will be minimal percentage-wise; and we can henceforth be sure that the daily summation is correct. The same time-decay will apply to the daily summation, in that after a week or so, the value of that specific datapoint won't matter. However if the state summations still don't match, people will question whether to use the data at all.

Thanks for that explanation, Dudzcom. Today I did some research using the Wayback Machine and found that the numbers were all correct through March 21. Georgia issues a report at both noon and 7pm, and there is a note in the table requesting that we use only the noon data for consistency. The Wayback Machine was missing the March 22 noon data but I found it in a Google cache of a local news report; it was 600 (the 7pm number was 620). So for consistency I updated the March 22 figure to 93 to give a total of 600 (and adjusted the US total accordingly), and also added the Georgia number for today (172) since the noon report is already out with a total of 772. Mark Taylor (talk) 19:38, 23 March 2020 (UTC)[reply]