Numbers and trends (6 April notes on COVID19, and a hurricane???)

Yes, a hurricane; it is still hurricane season in the southern hemisphere (technically a tropical cyclone, called a Typhoon in the area it is traversing).  SuperTyphoon Harold is moving through the islands northeast of Australia, and has killed 27 in the Solomon Islands. After a loop it is moving through Vanuatu and headed just south of Fiji over the next three days …

On the pandemic front, despite thousands of heartbreaking individual stories and much political drama, not much has changed over the weekend.  The progression seems to be pretty much on the lines forecast two weeks ago.  Here is the updated chart following various countries and states, along with statistical projections for for the outbreak:

The black line is a global estimate run what seems like years ago, on March 1st, that estimated a global death toll of 2.2 people per 10,000 of population.  The light grey line is a projection assuming COVID19 would have a similar death toll to the H3N2 influenza virus, but acting 5 times faster. That virus had a mortality rate of 2.96 per 10,000.  The green dashed line is a fit to the Italy data, and ends up around 3.25 per 10,000.  The red dashed line is the Spain data fit with an end mortality of 3.2 per 10,000, while the blue dashed line is a fit to New York data, also ends up at 3.25 per 10,000.  The points are actual reported deaths as of this morning.  Recall that death reports lag “cases” by 15 to 20 days.  I’m showing mortality rather than cases because the definition of “case” is so different between various jurisdictions, and is much more dependent on testing (which is a mess).

Most US states are at the lower end of the curve.  Here is a zoomed in view of that part of the chart, with just US data, and a couple more states added in:

When looking at “big data,” it’s vital not to be distracted by anomalies.  You need to stay focused on trends, and be very aware of how the data was collected and processed.  Weekend data is always suspect.  Take New Jersey (the blue diamonds).  Notice the very sharp change in the trend Sunday (the last marker).  Sign of hope, or reporting problem?  No way to know at this point – we will need several more days of data.  What about Washington state – it is well below the trend lines?  That is believable; they took early, aggressive action to limit the spread of the virus, and have decent enough testing protocols.  Georgia?  I don’t buy it – there are widespread reports of testing problems, and it is well known they are about at least a week behind everyone else in testing and reporting (tests are taking nearly two weeks to wind their way through the system).  Same with Florida and South Carolina.  So don’t get duped by the day to day wobbles in COVID19 reports (just as you shouldn’t get too excited by swings in hurricane forecast tracks!).  Watch the trends, take care of yourself and your family, follow the latest CDC guidelines, and don’t watch too much cable TV news!  If you have Amazon Prime, get caught up on Bosch – great series, new episodes coming out the 17th!

2 thoughts on “Numbers and trends (6 April notes on COVID19, and a hurricane???)

    • I wonder how many people have been introduced (or reintroduced) to jazz by that show (Bosch). I’m more of a Nordic/Symphonic Metal kind of guy (Nighwish, Delain, Amorphis, etc) but have enjoyed jazz in the past, Bosh reminded me of that.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.