How I monitor an area, or dealing with clutter.

I have been spending significant amounts of time examining data for colleges.  Only, I have very little visibility into what is happening directly on campus, just the schools dashboard.  Fortunately, the schools are located in fixed geographic regions.  For example, almost all Va Tech students live in Blacksburg in the Zipcode of 24060.  And, the state gives the following data by zip code:


The number of cases and testing are cumulative numbers.  But, by taking the time derivative (e.g., subtracting one day from the previous day), I can get the number of cases and tests per day.  And from that, I can derive a metric of percent positive.

For Va Tech, this looks like:

My overall approach is to look at the baseline number of cases (before the students arrive), compute the mean and standard deviation of the pre-arrival data, and track based on the number of standard deviations above the mean.

For Va Tech, the mean is low, so I will notice an outbreak when it is 3 standard deviations above the mean.  For 22040, that was at 8 cases a day, or 0.02% of the population.

In other places, there are either more pre-student arrival cases (noise) or non-student outbreaks (clutter).  As an example of clutter, I will show the cases for Longwood in Farmville:

The cases in Farmville around August 1 were unrelated to the university; they were the result of an outbreak at the Farmville ICE detention center.  What it means is, short of subtracting those cases from the data, the sensitivity of cases at Longwood is about 30-40 cases, or about 1% of the student body.  This is another way of says, I will not see an outbreak at Longwood until it is too late.

The following list highlights my sensitivity at the various schools:

     School     Community   Thresh  %of 
                Mean   Std.         students
     Va Tech    2.0    2.2    8.6   0.03
         GMU    4.5    2.9   13.3   0.04
         UVA    5.3    4.9   19.9   0.08
         ODU    3.0    2.7   11.0   0.04
         JMU    4.7    3.0   13.5   0.06
         CNU    3.7    2.4   10.9   0.22
         UMW    4.8    3.1   14.0   0.27
     Radford    2.2    2.5    9.7   0.12
         VCU    3.9    3.0   12.8   0.04
         W&M    1.5    1.1    4.8   0.06


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)