Talk:Statistics of the COVID-19 pandemic in India

Lower no. of daily tests on each sunday

Hi Shanze1 and Timbaaa, their is an intresting find from the Daily Tested samples chart. On each sunday, (from 3 May 2020) no. of daily tests significantly dips, as i was curious about this weekly affair, i tried to do some finding, although i don't have any source to support the claim but a near guess i found out is that on sunday private labs are closed and that dip is due to that. If i had a source than i would definatly want to add a note related to this in the chart, as than a reader can have a rough indication of how much the private labs are testing by comparing the tests of sunday with that of just a day before (i.e. saturdays). If you guys get any source, let me know. Mayankj429 (talk) 13:33, 12 June 2020 (UTC)

Mayankj429, Okay - Timbaaa -> ping me 15:02, 12 June 2020 (UTC)

Mayankj429, Sure Shanze1 (talk) 03:50, 13 June 2020 (UTC)

Does this chart add any value here?

  New cases as percentage of active cases   7-day moving compound average of new cases as percentage of active cases   7-day moving CAGR of active cases

Pinging Shanze1, Timbaaa, Mayankj429, Susam Pal. Just wanted to check with you guys about this chart. I've always felt it makes more sense to see the growth rate of new cases against active cases rather than the total confirmed cases. Also, it should be easy to see that the lower this percentage, the lower the growth rate of active cases.

Once this measure goes below a certain percentage (somewhere around 6% for India currently, it depends on how quickly we are discharging patients as per policy), the growth rate of active cases should go negative. -- Ashinpt (talk) 17:51, 3 July 2020 (UTC)

Ashinpt, I have not seen this type of chart used anywhere before. Could you link some where I can be read about it?(like [1] or something) If the chart should be added to the article, it definitely should accompany some explanation about the chart. Common reader may not know about it. - Timbaaa -> ping me 01:43, 4 July 2020 (UTC)

Timbaaa, I don't think I have seen it anywhere either. Not that I was going around looking for it. This has been created by me from the data that we have been gathering from MOHFW and putting on this page. There are 3 lines in the chart.

• The first one is New Cases as a percentage of the previous day's active cases. It is the active cases that contribute to the spread of the virus. It makes more sense to see the growth rate based on active cases, rather than confirmed cases. Just as an example. Let's say the total confirmed cases to date are 100,000.

Scenario 1: The active cases are 50,000. New cases next day number 5,000. Here, new cases as a percentage of active cases will be 10%.

Scenario 2: The active cases are only 500. New cases next day number 5,000. Here, new cases as a percentage of active cases will be 1000%.

Scenario 2 (which is admittedly an extremely unlikely scenario) should be more worrying, suggesting a new uncontrolled spurt of cases, even though in both scenarios the growth in the confirmed cases will be the same.

• The second one is where we take a compounded average over 7 days of the previous curve, just to smoothen the curve. This one involves a bit of calculation. So, hypothetically, if the the daily data for new cases as a percentage of active cases for 7 days were : 10%, 15%, 20%, 5%, 0%, 12% and 8%, the compounded rate would be c = ((1.10*1.15*1.20*1.05*1.00*1.12*1.08)^(1/7))-1 = 9.83%
• The third one is just the average growth rate of active cases over the previous 7 days. For e.g., if the number of active cases today is 10,000 and 7 days ago, it was 8,000, the CAGR would be ((10,000/8,000)^(1/7))-1 = 3.24%

The first two lines will always be non-negative, but the last line can be negative.

If the second line is consistently below a critical percentage (which looks to be 6-7% for India) for a sustained period of time, the number of active cases will start going down and the third line will go into negative territory. As an example, if the second line goes down to 4% consistently, the number of active cases will go down, as the number of recoveries will outpace the number of new cases. -- Ashinpt (talk) 06:39, 4 July 2020 (UTC)

Ashinpt, IMO, It can be added with explanation. But, I think WP:NOR might come in the way; I'm not exactly sure. Let's wait for what others have to say about it. - Timbaaa -> ping me 07:02, 4 July 2020 (UTC)

To add to discussion: COVID-19 pandemic in Turkey has similar chart. - Timbaaa -> ping me 07:06, 4 July 2020 (UTC)

Also the data for percentage of new infections over active cases seems to be wrong. Starting from 11-March-2020 the New cases as percentage of active cases, should be as follows:

y1 = 0.1754, 0.2, 0.1014, 0.0417, 0.2737, 0.0404, 0.1917, 0.1045, 0.1477, 0.2551, 0.3194, 0.1368, 0.2547, 0.1087, 0.1573, 0.139, 0.1872, 0.1026, 0.1176, 0.2032, 0.1179, 0.265, 0.1263, 0.2059, 0.1886, 0.1569, 0.1828, 0.1178, 0.1029, 0.1133, 0.1484, 0.1158, 0.1239, 0.1125, 0.1578, 0.1096, 0.0763, 0.0926, 0.0779, 0.0996, 0.108, 0.0879, 0.0937, 0.0736, 0.0978, 0.0786, 0.0979, 0.0692, 0.0724, 0.0789, 0.0754, 0.0698, 0.0909, 0.0886, 0.0867, 0.1212, 0.08, 0.0992, 0.0894, 0.0833, 0.079, 0.0957, 0.0783, 0.0742, 0.0756, 0.0772, 0.0749, 0.0924, 0.0931, 0.0845, 0.0918, 0.0882, 0.0918, 0.0956, 0.092, 0.0905, 0.081, 0.0769, 0.0763, 0.083, 0.0922, 0.0931, 0.0899, 0.0837, 0.0878, 0.0872, 0.0888, 0.0853, 0.0828, 0.0799, 0.0769, 0.0747, 0.0727, 0.0772, 0.0786, 0.0799, 0.0751, 0.0696, 0.0707, 0.0821, 0.0813, 0.0863, 0.091, 0.085, 0.0839, 0.0872, 0.0907, 0.0913, 0.094, 0.098, 0.0926, 0.0889, 0.0817, 0.0844, 0.0919, 0.0925, 0.1056, 0.0962, 0.0855, 0.0878, 0.0939, 0.0917, 0.0957, 0.0979, 0.0952, 0.0915, 0.092, 0.0987, 0.1021, 0.0973, 0.1042, 0.1035, 0.0923, 0.0918, 0.1073, 0.112, 0.1073, 0.104, 0.1029, 0.096, 0.0952, 0.0987, 0.101, 0.1011, 0.0964, 0.0914, 0.0888, 0.0896, 0.0945, 0.103, 0.0994, 0.1024, 0.0977, 0.0838, 0.0947, 0.1025, 0.0976

Please cross-check and update. e.g. On 11-Mar-2020, new infections were 10 and active cases were 57, so it comes to 0.1754. On timeline, the relevant data is
2020-03-10;0;3;50
2020-03-11;0;3;60
This means 3 recoveries till 11-Mar-2020.--- Pankajsach (talk) 05:23, 14 August 2020 (UTC)

Courtesy ping: Ashinpt - Timbaaa -> ping me 07:10, 15 August 2020 (UTC)

The calculation is new infections on 11-Mar-2020 (10) divided by active cases at the end of the previous day 10-Mar-2020 (47). This should give you a value of 0.2128 or 21.28% for 11-Mar-2020. Hope this clarifies. -- Ashinpt (talk) 18:58, 16 August 2020 (UTC)

The same chart for Kerala as an example

  New cases as percentage of active cases   7-day moving compound average of new cases as percentage of active cases   7-day moving CAGR of active cases

Just to illustrate what happens when the new cases as a percentage of active cases is low, the above chart is for Kerala.

Notice that from the second week of April till the middle of May, the 7 day CAGR of the active cases is mostly in the negative. This corresponds to a period where the number of active cases in Kerala declined from somewhere close to 260 all the way down to 16. As you can see, the second (magenta) line stays pretty low during this time, sometimes dipping below 2%.

In recent days, Kerala has not managed to consistently go below the 6% mark, as it has less control over how many cases it imports from out of state. I believe 85-90% of the cases in the state so far have been imported from outside. As a result, the number of active cases has been mostly increasing recently.-- Ashinpt (talk) 19:21, 4 July 2020 (UTC)

Ashinpt, Seems no one is participating, go ahead and add the chart. - Timbaaa -> ping me 02:20, 7 July 2020 (UTC)

Case fatality rate graph

This extends into August for some reason, I can't see why. All the best: Rich Farmbrough 01:27, 29 July 2020 (UTC).

Thanks for notifying. It seems to be a bug in the module:graph itself.- Timbaaa -> ping me 11:59, 29 July 2020 (UTC)

Same for daily new cases vs active cases graph.- Ashinpt (talk) 12:27, 29 July 2020 (UTC)

Ashinpt, I previewed some scenarios, following one worked(it's a bodge not a solution):

• Adding non-null x and y data fields(at the beginning or at the end) to the actual fields for 7 more days(approx.), will fix the scaling problem. i.e., in 7 days, it will be fixed on its own.

The proper solution has to be worked on module:graph- Timbaaa -> ping me 14:47, 1 August 2020 (UTC)

@Ashinpt and Rich Farmbrough: Letting you know that, it is fixed in the sandbox version. Will be updated to main module soon.- Timbaaa -> ping me 02:56, 8 August 2020 (UTC)

Very well done! All the best: Rich Farmbrough 16:03, 10 August 2020 (UTC).