About this blog

The goal of my reports are:

  •  provide a clear and concise status of COVID in our community
  • Apply physics-based data analysis techniques to identify and understand the trends.
  • Provide straightforward graphical representation of the data
  • Capture the uncertainties in the data.
  • Present without politics, fear uncertainty and doubt.

The differences in approaches are religious arguments between scientists (e.g., people looking at the physical reasons for something happening) and statisticians (e.g., data scientists), who parse the data looking for patterns.

To me, what is interesting (I am a physicist) is to try to deduce what is changing in the underlying mechanisms….In my case, we are we going from stable to exponential growth.  

In the data approach, the data is everything.  You might assume an underlying model, but you do not know anything about things like mask ordinances.  The advantage of that is you can automate it — run numbers for every county, and possibly every zip code in the country.

The down side is with the automated approach the the script has no situational awareness.  As an example, I see what is happening around here.  I see that some people are not following the ordinances, and compliance is less than it was in June.  The data shows that cases are increasing, but slowly.  

(In the end, what I am really trying to do is get a good estimate of Rt based on the current situation, and to see if things are changing).

I am not an epidemiologist.  I started this effort on March 10th during a conference call at work,  (it was a very redundant call).  The daily data had just been updated.  I wanted to see what was happening, how bad it would get. (IT WAS SCARY).  After that, I started tracking for my own interest.  Mostly, I was not happy with the information we were all getting.  I wanted to know what the data was showing:  R0, etc.  This was before the lockdown.  I started archiving the data each day (the state was not doing it at that time), and looking at the growth rates.  

As the lock down began, the curve was not flattening.  I was wondering if it would.  I started looking at it statistically.  When I noticed the number of cases in Virginia began falling below the trend line (linear line in a log plot), I started looking at the statistical significance: are we seeing random variations, or something meaningful?  And that is when I wrote my first blog.  
April 1:  Curve is within 2 sigma of the trend
April 8:  Every day in the last week is two sigma below the trend:  curve flattened.

I haven learned a lot about this:  two weeks from an event occurring until it shows up in the data.  

I started sharing to the broader Vienna Community  in mid-may, when I had zip code data.  That was around our peak, and I felt the need/responsibility to alert the community that it is bad out there (20-30 people per day getting sick in 22180).  

I have also adjusted my reporting as the data has improved.

I do not have a specific methodology.  I have not really documented my methods, as they are not that unique.  I am just applying it to the local area in more detail.  


Popular posts from this blog

Daily Status, August 29

Daily Status, Nov 8: Every trend I was tracking 10 days ago remains. Which is not good.

October 24: Vienna is not doing well, Virginia is not doing well, the USA is not doing well (but I will not talk about the latter)