jump to navigation

About Rolling Averages in Time Series Data March 29, 2020

Posted by Peter Varhol in Uncategorized.
Tags: ,

Okay, you can stop yawning now.  This is about coronavirus.  I have heard multiple claims today that coronavirus is “only” killing one percent of the people who have been infected in the United States.  Which is less than the flu kills annually.  So it can’t be as bad, and the media is dramatizing it.

This is grossly misleading enough so as to constitute a baldfaced lie.  Let’s examine why.

The fatality rate of the flu (whichever strain, and it varies) is a relatively static number, measured over a period of years.

The fatality rate of coronavirus, by contrast, is a moving average.  We haven’t yet completed a single season of the infection; in fact, probably less than three months.

There is a time lag between when a person is infected, and when it is fatal.  As near as I can tell, that time period is usually somewhere between two and six weeks, although there are also outlyers.  But what this means is that comparing those who are infected to date, with those who have died to date, is grossly underestimating the fatality rate.

And I’m not even looking at the millions of people who will likely be infected in the future months.  Further, we test and diagnose many more people for the flu than we are currently doing for coronavirus.  We are so far behind in testing that we really don’t know how many are infected.  That was why West Virginia had no cases until recently; they weren’t testing anyone.

To be fair, I’m not an epidemiologist by any means.  I was trained as a mathematician and statistician, however, and have a solid grounding in the experimental method.

Are we really so ignorant as to compare ongoing time series data to established (and static) averages over the years?  That seems to be what is happening.  And that’s equivalent to comparing apples with red wine.  Our infection growth rate, and fatality growth rate, are still growing on an exponential curve.

Pay attention, folks.  And stay safe.


1. Five Blogs – 31 March 2020 – 5blogs - March 31, 2020

[…] About Rolling Averages in Time Series Data Written by: Peter Varhol […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: