 |
| Data Source: https://www.worldometers.info/coronavirus/ Click on the image to enlarge |
Current Model
On 3/10, we had a total of 994 cases and 30 deaths, rapidly progressing to 19,367 and 255 on 3/20, nearly quadrupling to 68,211 and 1,027 in mere five days (3/25), and then exponentiating to 244,877 and 6,070 on 4/2.
Though the overall slope is nowhere close to being exponential (the exponential fit shows a moderate r-squared of 0.711, hence the near-perfect polynomial fit), it has recently transformed into an exponential slope (the graph below), meaning the death rate (death to total) has been accelerating.
The exponential regression fit is therefore producing a near-perfect r-squared of 0.990. In other words, the death rate has escalated from 1.51% on 3/25 to 2.48% on 4/2 -- an exponential jump indeed. Terribly sad!
Predictive Model
In order to develop a meaningful predictive model for the US, an auto-regressive "Roll Rate" needs to be developed or simulated off of the Serious/Critical cases (being published daily), meaning we need to develop an adjustable add-on rate to be added on to the front-end of the death rate given the fast-shifting change in the slope.
To be clear, we cannot apply the European model (mostly seniors) to predict the US rate as we have more of a hybrid outbreak here (40% young adults). Ideally, if the Serious/Critical group could be split up, the Critical cases may lead to better predictive/navigational add-on rates.
Stay safe!
-Sid Som
homequant@gmail.com
Data Source: https://www.worldometers.info/coronavirus/
Also Read: Regression Analysis of Global Coronavirus Stats – March, 2020
No comments:
Post a Comment