Analyzing Number of New Cases confirmed and Deaths Reported on a Given day due to the Outbreak COVID-19

I’m trying to find the pattern in which the number of cases and deaths due to corona virus increase on a day to day basis. With the very basic tools I’m equipped with, I have come up with some idea of death ratio and predicted the number of deaths using the number of confirmed new cases reported.

0.064589 60830 10 39289.4887
0.074559 48461 10 36132.03699
0.0860628 43744 15 56470.96685
0.100370429 41371 15 62286.37527
0.117487231 32480 15 57239.77894
0.134258417 29429 15 59266.36431
0.158891 30648 10 48696.91368
0.1862899 26158 10 48729.71204

Total 408111.6368, Average 4081.116

0.055794 64502 10 35988.24588
0.064589 60830 10 39289.4887
0.074559 48461 15 54198.05549
0.0860628 43744 15 56470.96685
0.100370429 41371 15 62286.37527
0.117487231 32480 15 57239.77894
0.134258417 29429 10 39510.90954
0.158891 30648 10 48696.91368

Total 393680.7343 Average 3936.807

I think these figures stand as upper bound and lower bound respectively.

So, the Number of deaths on 29th March would be between 3936 and 4081. My Prediction would be their mean 4008.

Interesting. Could you provide some more details on your data (e.g. the source, what do the columns mean,…)?


I get my raw data from this site

Thank you so much for responding to my answer Sir.

First of all, the basic idea is to try and use the notion of Death ratio. For an ongoing epidemic, death ratio could be calculated using the formula

CFR(Crude Fatality Ratio) = Y/f(X- D) where Y is the No. of deaths on a given day, X be the Date we consider, D be the average number of days from confirmation of disease to death. And f(x) be the number of new cases reported on the day x.
If this is guess, just by approximation, we can see taking
D = 4, we get CFR = 0.085709 (This I calculated by taking the mean of CFR from the day March 9).

Now using this CFR, we can approximate the number of deaths today(30th March) by considering the number of new confirmed cases four days ago, that is of 26th of March.

That would be Y = CFR * f(X-4) = 0.085709 * 60830 = 5213.68.

Using D = 3, we will have Y = 4801

Using D = 2, we have Y = 4287

Using D = 1, we have Y = 3248.

By giving them different weightages, we can probably predict the lower bounds and upper bounds of number of deaths to be reported tomorrow. Of course it would be between 3248 and 5213. So, expecting it to be around 4200.

The data in the site is only those reported, so we cannot expect that to be exact (though they produce true facts and figures as far as they can fetch information, they are limited by the response of authorities towards the pandemic).

I have literally done the same thing back in my first answer also, where the first column would mean the CFR calculated for D =2,3,…8,9. And I have given the weightage of 15 each for the four central days and 10 each for the four boundary days.
But that analysis do have its limitation I understand now. Because, Most of the authorities make use of the test only after severe and extreme symptoms are observed. In that case, the number of days from confirmation to day of death decreases considerably. As far as we are concerned, D becomes lower. I threw D into the range of 5.5 yesterday, but as I describe today, I have conformed to a D near 2.5 with almost equal distribution around for four days.

The second column is the number of new cases confirmed exactly D days before the date to which we are approximating the number of deaths. The third column corresponds to the weightage I gave, and the fourth column is their product with their sum as 100 times the value of my guess.

Don’t be mistaken to find from the site that only 3115 new deaths have been reported yesterday. Since yesterday was sunday, the number has much to do with the reporting of data more than the reduced impact of the virus. It is also clear from the confirmation of less than 60,000 new cases, which is also a low among the last four days.

If this scenario persists, then there could be an argument made that the pandemic has stopped going exponential. But I feel its just starting to heat up and the numbers are going to bounce.

Hope I made myself clear with my data and methods of analysis. I am not so competent or equipped to do a more scientific one. But at least I can come up with a rough idea of the spreading of the virus so that ultimately I can know the curves when and where this is going to reach its peak and start subduing.

Thank you once again.