Thread by @tjhaywood7

Not placing blame. Not accusing anyone of purposeful deception. Just pointing out a problem in the hope it can be fixed. I am so thankful for all the data being provided. I just think it may be time for a little clean up.
#covid19 #SARSCoV2 #florida @HealthyFla

The data being provided for analysis by FL DOH in their Data Hub (open-fdoh.hub.arcgis.com/search?q=covid) does not match the data in the FL DOH COVID-19 Dashboard (experience.arcgis.com/experience/96d…) and different variables within the FL DOH COVID-19 Dashboard contradict each other.

Furthermore, Florida is grouping people into three groups: FL residents in FL, non-FL residents in FL, and FL residents outside of Florida. However, the various measures on the FL DOH Dashboard alternately separate, combine, and/or leave off these groups in calculations.

For instance:
 
Palm Beach County in the FL DOH Dashboard today (5/22/0202 afternoon):
Total People Tested: 6,100
Positive: 5,072
Negative: 54,546
Positive Residents: 4,877
Positive Non-Residents: 93
Hospitalizations: 1,021
Deaths: 308

Data Hub
CasesAll:5072
C_NotFLRes:93
C_AllResTypes:4979
C_FLResOut:2
C_FLRes:4977
Deaths:311
C_FLResDeaths:311
C_NonResDeaths:11
TPositive:5072
TInconc:226
TPending:110
TNegative:48897
TNegNotFLRes:533
TNegRes:55058
T_negative:55591
T_positive:5072
T_Total_Res:53343
T Total:61100

Furthermore, the data dictionary definitions do not actually always match the way the variables are being used. (The data dictionary can be accessed by going to the dashboard lard, clicking the hamburger lines in the top right, and choosing “About This Map”.)

As you can see, TPositive is supposed to be, for some strange reason, "Number of PUIs with test results, including negative, positive and inconclusive, but excluding pending or awaiting testing. This is the total number of people with test results in our system".

But it's actually giving total positives for all residents (in and outside Florida) and non-residents, which actually matches the value in T_positive (which has a definition of "Testing: Total positive persons tested for all Florida and non-Florida residents").

T_positive does match CasesAll, which has a definition of "Cases only: Total (sum) of all confirmed positive COVID-19 people in Florida or who are Florida residents.”

While this value does match the value under “Positive” on the Florida Testing Tab for Palm Beach, this “Positive” total does not even match total of “Positive Residents” plus “Positive Non-Residents” on the Cases by County Tab for Palm Beach.

T_NegRes is "Testing: Total number of negative Florida Residents tested [presumably inside and outside of Florida]" and T_NegNotFLRes is "Testing: Total number of negative non-Florida residents in Florida tested".

Adding them together does equal t_negative, defined as "Testing: Total negative persons tested for all Florida and non-Florida residents". However, this does not match TNegative, which has a definition of "PUI: Number of PUIs with negative test results (quality control group)".

However, neither TNegative nor t_negative matches the value under “Negative” on the Florida Testing Tab for Palm Beach.

On the FL Testing Tab for Palm Beach, the total number of people tested matches the data in t_total, which actually isn't defined in that data dictionary.

But I can't find a combination of t_negative, t_positive, TNegative, TPositive, TInconc ("PUI: Number of PUIs with inconclusive test results (quality control group) ), and/or TPending (PUI: Number of PUIs with test results pending (quality control group)) equal to t_total.

And the sum of “Positive” and “Negative” on the Florida Testing Tab for Palm Beach don’t even match the value of “Total People Tested” on the same Tab. Adding Pending Tests and Inconclusive Tests (data both taken from the FL DOH Data Hub) still doesn’t account for the difference.

On top of all of this, the FL DOH dashboard only lists hospitalization of FL residents, which matches C_HospYes_Res, but it's not stated clearly if this is FL residents within FL or anywhere.

Finally, even though not shown on the dashboard though, C_HospYes_NonRes lists hospitalized non-FL residents so I can combine the two to arrive at the hospitalization stat for Florida that are listed on the Johns Hopkins dashboard.

If I have miscalculated the data, please someone let me know. But I just can’t figure out what’s causing these problems. It may just be time for some data clean-up and some clarification added to the dashboard itself. #covid19 #SARSCoV2 #Florida @HealthyFla

See thread.

#covid19data
#Florida
#SARSCoV2
#covid19

@HealthyFla #covid19 #SARSCoV2 #Florida #covid19data today (5/25/20) the data does look a bit better, at least for Palm Beach County. T_negative and t_positive now match the “Positive” and “Negative” fields on the Florida Testing Tab for Palm Beach. However, 2 problems still...

TNegative still does not match t_negative for Palm Peach County and T_Total_Res (supposed to be total TESTS of residents) is smaller than even T_NegRes (total negative residents) which makes no sense...

No combination of variables adds up to T_total, which, despite not being in the data dictionary, matches “Total People Tested” on the Florida Testing Tab for Palm Beach County.

TPositive + TNegative + TInconc + TPending = PUIs Total do that could explain TNegative but then why are only a subset of negative people included under PUIs in Florida?