This is a very short follow-up piece on my earlier article(Math of the Macabre) that I had written on April 4th with data till the end of April 3rd. Please refer to the previous piece before reading this.
A quick summary: I had developed an equation of the form
y = 0.0337x4 + 0.7715x3 + 5.0925x2 + 94.975x + 3296.3
Over the past 6-7 days, I have been comparing the number of deaths as predicted by my equation and the actual number of deaths that indeed happened to see the % error. The primary reason behind doing this is to get a more concrete estimate of the COVID-19 pandemic at hand. Can the death rate still be modeled by my equation? Does my equation underpredict? If yes, then where are we going wrong in terms of containing the spread? If no, then how can we increase the magnitude of the overprediction? Here’s the answer to my questions.
I am happy to report that my equation is not accurate anymore (ironical but desirable in this situation). Up until the 8th of April, the % by which my equation was unpredicting kept on decreasing gradually. However, after the 9th of April, the equation has started over predicting the number of deaths! This means that the slope of the equation of the number of deaths (real-life) has started tapering! Although this is not a lot, it does serve as an assurance that the pandemic isn’t a maliciously upward curve! I know this is too early to celebrate, but every ray of hope matters in times like these : )
|Date||Days elapsed (x)||Predicted||Actual||Error|
|25 March 2020||20||18,797||21,282||-13.2%|
|01 April 2020||27||42,668||47,192||-10.6%|
|05 April 2020||31||65,241||69,427||-6.4%|
|06 April 2020||32||72,168||74,660||-3.5%|
|07 April 2020||33||79,667||82,043||-3.0%|
|08 April 2020||34||87,770||88,457||-0.8%|
|09 April 2020||35||96,508||95,694||0.8%|
|10 April 2020||36||1,05,913||1,02,687||3.0%|