Copyrights @ Journal 2014 - Designed By Templateism - SEO Plugin by MyBloggerLab

, ,

Infovis shenanigans

I have been wanting to do this for days. It is my profession, after all. Or rather, it is a very important part of my profession - information visualization. And I haven't pored over my Edward Tufte and my Chaomei Chen for nothing for all these years. Only reason I left it so long was that I needed a substantial number of days of both covid 19 test numbers as well as test results to be able to show what I want to show. So, here without further ado, it is:

What you see in this first graph is what you see everywhere from the worldometer site to every other imaginable platform. In fact it is the only thing one seems to see. It is a graph that is made out of accumulated numbers. Meaning that the numbers of the current day get added onto those of all the previous days. Naturally, the graph keeps moving up and up as more and more days during which more and more tests are conducted are added; given that it takes a while, possibly weeks before the cases diagnosed have a chance to recover and can then be subtracted from the total. Which will happen eventually but not right away. So, there will be no downward turn in this graph anytime soon. In fact, heaven forbid a sudden downward turn  - it would mean nothing but people suddenly dying in huge numbers overnight.

It is only natural that people who see graphs like this completely freak out about the unstoppable spread of the pandemic. Completely understandable. But a misplaced fear nevertheless. (As the saying goes - "trust me, I am a designer" ;-). But all kidding aside, this really is my job. I have been doing it and teaching it for nigh on 45 years and I have yet to be fired for incompetence.

Now look at this second graph. It has been compiled out of the exact same numbers that are on the first graph. The difference is that these are percentages that I have calculated out of tests conducted per day and cases identified out of that number. It rises and dips, but the bottom line is that overall it is even. Ergo, no horrifying increase in disease spread as the first chart would have you believe. 

Again folks - you are looking at the exact same data, except that it can be shown to you in very different ways. As any info-vis educated graphic designer worth his or her salt can show you in a flash. Which is what I have done here. 

Which I have not done to soothe your frayed nerves. I have done this to get you to finally turn away from all this and to start looking at the real elephant in the room which is the collapse of the world economy. 

Note: The health ministry gives out test numbers and case numbers here everyday on a web page that is actually very easy to understand. There is a lot of suspicion in the Turkish public that they are hiding the numbers, that these results (both tests and cases) are not accurate. Now, I would be the last person to trust any government agency anywhere in the world, much less my own. So, I am definitely taking this into account. But it is what I have to go on. I have looked for concise, easy to figure out portals like the one here for other countries (well only the US and the UK to be honest) but I can't seem to find one. Would be nice to be able to do this exercise with some other data sets as well. Especially to find out if the guys here are fibbing or not. But alas...

.........................

Here is a list of all the days that I have been keeping, in case you are wondering how things are going since April 6th, which is the last day on the graphs above:
  • 19 March (1981 tests, 168 cases) % 8,48;
  • 20 March (3656 tests, 311 cases) % 8,50; 
  • 21 March (2953 tests, 277 cases) % 9,38; 
  • 22 March (On this date the case number was given (289) however the total test number was not given by the Minister. Some people on twitter came up with a very low test number (1775) which gives a very high percentage result of course - %16,46. This number of 1775 tests, looking at the percentages trend from the days before and after is very likely to be completely bogus, but I am adding it in anyway.)
  • 23 March (3672 tests, 293 cases) % 8.2
  • 24 March (3952 tests, 343 cases) % 8.67
  • 25 March (5.035 tests, 561 cases) % 11.1 (yes, today it has gone up. Taking note)
  • 26 March (7.286 tests, 1.196 cases) % 16.4 (so yes, this is a big jump. Taking note, of course)
  • 27 March (7533 tests, 2069 cases) % 27.4 (This does seem to be spreading very quickly. So could it be of more concern than I think? _____ Or, having thought about this for a bit: It could also be the spread pattern of any seasonal flu, of any corona virus. Starts slow then increases rapidly. No way of knowing that. At least for me.) 
  • 28 March (7641 tests, 1704 cases) % 22.3, 15 deaths
  • 29 March (9982 tests, 1805 cases) % 18.1, 16 deaths
  • 30 March (11535 tests, 1610 cases) % 13.9, 33 deaths
  • 31 March (15422 tests, 2704 cases) % 17.5, 23 deaths
  • 1 April (14396 tests, 2148 cases) % 14.9, 37 deaths
  • 2 April (18757 tests, 2456 cases) %13.1, 46 deaths
  • 3 April (16160 tests,   2786 cases)  %17.2, 144 deaths
  • 4 April (19664 tests, 3013 cases) %15.3, 67 deaths
  • 5 April (20065 tests, 3135 cases) %15.6, 76 deaths
  • 6 April (21400 tests, 3148 cases) %14.7, 73 deaths
  • 7 April (20023 tests, 3892 cases) %19.4, 75 deaths
  • 8 April (24900 tests, 4117 cases) %16.5, 76 deaths
  • 9 April (28578 tests, 4056 cases) %14.9, 87 deaths
  • 10 April (30864 tests, 4747 cases) %15.3, 96 deaths
  • 11 April (33170 tests, 5138 cases) %15.4, 95 deaths
  • 12 April (35720 tests, 4789 cases) %13.4, 97 deaths
  • 13 April (34456 tests, 4093 cases) %11.8, 98 deaths
  • 14 April (33070 tests, 4062 cases) %12.2, 107 deaths
  • 15 April (34090 tests, 4281 cases) %12.5, 115 deaths
  • 16 April (40427 tests, 4801 cases) %11.8, 125 deaths
  • 17 April (40270 tests, 4353 cases) %10.8, 126 deaths
  • 18 April (40520 tests, 3783 cases) %9.3, 121 deaths
  • 19 April (35344 tests, 3977 cases) %11.2, 127 deaths
  • 20 April (39703 tests, 4674 cases) %11.7, 123 deaths
  • 21 April (39429 tests, 4611 cases) %,11.6, 119 deaths
  • 22 April (37535 tests, 3083 cases) %8.2, 117 deaths
  • 23 April (40962 tests, 3116 cases) %7.6, 115 deaths
  • 24 April (38351 tests, 3122 cases) %8.1, 109 deaths
  • 25 April (38308 tests, 2861 cases) %7.4, 106 deaths
  • 26 April (30177 tests, 2357 cases) %7.8, 99 deaths
  • 27 April (20143 tests, 2131 cases) %10.5, 95 deaths
  • 28 April (29230 tests, 2392 cases) %8.1, 92 deaths
  • 29 April (43498 tests, 2936 cases) %6.7, 89 deaths
  • 30 April (42004 tests, 2615 cases) %6.2, 93 deaths
  • 1 May (41431 tests, 2188 cases) %5.2, 84 deaths
  • 2 May (36318 tests, 1983 cases) %5.4, 78 deaths
  • 3 May (24001 tests, 1670 cases) %6.9, 61 deaths
  • 4 May (36771 tests, 1614 cases) %4.5, 64 deaths
  • 5 May (33283 tests, 1832 cases) %5.5, 59 deaths
  • 6 May (30303 tests, 2253 cases) %7.4, 64 deaths
  • 7 May (30395 tests, 1977 cases) %6.5, 57 deaths
  • 8 May (33687 tests, 1848 cases) %5.4, 48 deaths
  • 9 May (35605 tests, 1546 cases), %4.3, 50 deaths
  • 10 May (36187 tests, 1542 cases) %4.2, 47 deaths
  • 11 May (32722 tests, 1114 cases) %3.4, 55 deaths
  • 12 May (37351 tests, 1704 cases) %4.5, 53 deaths
  • 13 May (33332 tests, 1639 cases) %4.9, 58 deaths
  • 14 May (34821 tests, 1635 cases) %4.7, 55 deaths
  • 15 May (38565 tests, 1708 cases) %4.4, 48 deaths
  • 16 May (42236 tests, 1610 cases) %3.8, 41 deaths
  • 17 May (35369 tests, 1368 cases) %3.8, 44 deaths
  • 18 May (25141 tests, 1158 cases) %4.6, 31 deaths
  • 19 May (25382 tests, 1022 cases) %4, 28 deaths
  • 20 May (20838 tests, 972 cases) %4.6, 23 deaths
  • 21 May (33633 tests, 961 cases) %2.8, 27 deaths
  • 22 May (37507 tests, 952 cases) %2.5, 27 deaths
  • 23 May (40178 tests, 1186 cases) %2.9, 32 deaths
  • 24 May (24589 tests, 1141 cases) %4.5, 32 deaths
  • 25 May (21492 tests, 987 cases) %4.5, 29 deaths
  • 26 May (19853 tests, 948 cases) %4.7, 28 deaths (I am beginning not to trust the numbers of these past 3 days since they fly in the face of what Knut Wittkowski and other epidemiologists have told over and over. Once a curve starts going down, it doesn't jump back up, according to them. "Nature never jumps, it moves very smoothly" was what Wittkowski said, in fact. Something is off here, unless it is some kind of human error. But 3 days in a row? Are they trying to delay going back to normal?) 
  • 27 May (21043 tests, 1035 cases) %4.9, 34 deaths (Yep. I am definitely smelling a rat. According to this percentage we are stuck since pretty much May 8th around the same place, with a 3 day long dip to around 2.5 somewhere in the middle. This feels very wrong to me. Very.)
  • 28 May (3359 tests, 1182 cases) %3.5, 30 deaths
  • 29 May (36155 tests, 1141 cases) %3.1, 28 deaths (Back to where it is supposed to be. So, what were the 4 days of %4+ numbers all about then? Human error?)
  • 30 May (39230 tests, 983 cases) %2.5, 26 deaths
  • 31 May (35600 tests, 839 cases) %2.3, 25 deaths
  • 1 June (31525 tests, 827 cases) %2.6, 23 deaths (This is the date that Turkey has completely opened its economy back up. Including regular cafe, tavern etc service. No "new normal" in other words, back to the old normal. I am going to keep track from now to see whether this makes any difference to the graph, which if what the fear mongers out there turn out to be correct in their gloomy predictions should come about in about 15 to 20 days or so. And yes, we have our share of highly effective doomsayer mainstream media and opinion leaders and journalists, as do all other countries.)
  • 2 June (32325 tests, 786 cases) %2.4, 22 deaths


2 comments: