Monday, May 25, 2020

C19 Update with Brazil and US bug fix

You might have noticed that my total C19 death count for the US is a bit higher than seen on the Johns Hopkins site or national headlines. Pulling on the Johns Hopkins GitHub data I showed US passing 100K deaths on 23 May. This was getting progressively more annoying, so I checked the python code and printed the tail of my pandas data frame at key points. In one line there is a search to find CountryRegion=US, or that is what it was supposed to be doing. Looking at the tail showed the code was keeping and summing up every country and region with 'us' somewhere in the name (Australia, Austria, ..., Mauritius, Russia, US). This group is dominated by the US, but adds a few thousand non-US deaths to all my previous plots. The bug fix was easy; now the data frame search to find US includes the parameter 'case=True' so that only all caps US is snagged. Oh well. From this post on, the US count will be correct.

Brazil is much in the news and is today's guest region. Too early for lognormal fit, but maybe we will stick with it for a few posts and see when it rolls over.

Total US C19 deaths 97,720
Future estimates
   105132 +/- 5648  =  Jun 1 total (95% confidence)
   124946 +/- 5648  =  Jul 1
   134794 +/- 5648  =  Aug 1
   139677 +/- 5648  =  Sep 1

Total world C19 deaths 345,059
Future estimates
   362923 +/- 27998  =  Jun 1 total (95% confidence)
   419027 +/- 27998  =  Jul 1
   442964 +/- 27998  =  Aug 1
   452735 +/- 27998  =  Sep 1

Brazil: Current total = 22666

No comments: