One of the first effects of the COVID-19 pandemic to cancel the most fun event in American sports: the NCAA men’s basketball tournament. It’s nicknamed “March Madness” because of, well, the significant about of sheer craziness involved.
From first-round upsets to the championship game, it’s four weeks of constant excitement. If you don’t believe me, check out this highlight reel of the 2019 NCAA tournament (long but worth it!).
And then in 2020, it was gone.
Just four days before the selection committee was set to announce the 68 teams who qualified, right in the middle of a conference tournament game, the National Collegiate Athletic Association (NCAA) cancelled the whole thing. And while professional sports leagues are planning to finish out their seasons (AFL season restarts tomorrow!), college basketball’s seniors have already graduated and moved on. So this year’s tournament is definitely gone for good.
While certainly not as big a deal as the 4,619 people who had died as of then (and it’s nearly 100 times that now!), the tournament’s cancellation still came as a disappointment to millions of college basketball fans. Including me.
My March Madness
I decided that, if we can’t have March Madness in reality, we can at least have it in our imaginations. And so I picked a field of 68 team, taking into account all the factors that the real selection committee does: record, ranking, strength of schedule, and conference tournament results.
And here is the bracket:
The official 2020 NCAA imaginary tournament bracket (click for a larger version)
I then simulated the entire tournament using whatifsports.com’s college basketball simulator. I announced the results, in real time, at my appropriately-named Twitter account, @fixthemadness.
Results
As each round of the tournament completed, I reviewed the results as a post on this blog. Read the posts linked in the list below to see how it all played out, round-by-round:
…except that as the COVID-19 crisis worsened, I turned my attention to that and never posted the Final Four results. Here they are. The Final Four consisted of the winners of each of the four regions (East, South, Midwest, and West). Both Final Four games were played on Saturday, April 4th – the first at 6:07 PM ET and the second at 8:07 PM ET.
The results were (click on the score to see the box score and play-by-play from whatifsports.com):
That set up an exciting matchup between the historical powerhouse team, the Duke Blue Devils (#1 seed in the South), and the surprising Pirates of East Tennessee State.
The game was close throughout the first half and early in the second, but then Duke pulled away. The final score (with link to the box score and play-by-play):
Congratulations to the imaginary Duke Blue Devils on winning the imaginary 2020 NCAA men’s basketball championship!
The full bracket
Here is the full bracket with all results, leading up to imaginary Duke’s imaginarily historic imaginary win:
All results from the imaginary 2020 NCAA men’s basketball tournament (click for a larger image)
What’s next?
I had a blast doing this. It went on for a month, and then it was over, and @fixthemadness wasn’t ready for it to be over. What could I possibly do for an encore?
Find out Friday!
Also, coming tonight: today’s daily update of COVID-19 cases and deaths.
Graphs day 84, pandemic day 89, day 160 since the first cases were diagnosed.
Today is the day that Earth passed seven million cases of COVID-19, eight days before my mid-May prediction said we would. The good news is that the number of deaths continues to run behind those predictions.
Total cases of COVID-19 diagnosed worldwide: 7,069,278
Total deaths: 405,587
I will continue to show cases using the new methods I explained yesterday: plotting the number of newly reported cases, resulting in a curve that goes up and down as the pandemic waxes and wanes. Here’s what that type of graph looks like for the entire world over the entire history of the pandemic:
Cases of COVID-19 reported per day from the beginning. The blue line shows the actual data; the red line shows the smoothed data, which shows overall trends without the daily variations. Click for a larger version.
As you can see from the graph, we are very much in the middle of this pandemic.
Cases per million people by country
Yesterday I divided countries into four categories:
Countries that quickly contained their COVID-19 epidemics, and never had a large number of cases per million people
Countries that had many cases, but have since gotten their epidemics under control and have few new cases
Countries where a significant number of cases are still being diagnosed, but fewer and fewer each day – these countries are headed in the right direction
Countries where the epidemic is still getting worse
I’ll focus primarily on the last two categories – improving and getting worse. And unfortunately, we’ve already had one country move into the “getting worse” column.
First, the countries where the epidemic seems to be getting better, or at least staying the same:
Countries where newly-reported cases per million people are steady or decreasing
Notice that Peru is gone from the graph – the situation in Peru is getting worse again. Belarus, the United States, and Russia all make me nervous, because it looks like the curves could start going up again at any moment.
And now the countries where the epidemic continues to get worse, now sadly including Peru:
Cases per million people reported yesterday in Brazil and Sweden were almost exactly the same, but Sweden’s rate is a little higher. Sweden is starting to look really bad.
and as a graph comparing case rates on an even scale from each country’s “day zero,” the date on which cases reached 1 in 1,000,000:
Deaths per million people by country
When it comes to death, the dead stay dead, so it makes sense to consider cumulative deaths, resulting in curves that steadily increase with time. Those curves, for the Big Ten countries:
Deaths per million people in various countries (click for a larger image)
Note the large spike in deaths per million in Chile, presumably due to reporting of deaths catching up with the actual data.
Want to try out some other ways for yourself? You can get the data that I used to make these graphs from the European Centers for Disease Control’s Coronavirus Source Data; choose “all four metrics.” You’re still welcome to use my Excel template – I added a section for making the deaths graph to what is now verison 4.1, so that is the only one you need.
Update tomorrow, and every day after that until this pandemic comes to an end.
Graphs day 83, pandemic day 89, day 160 since the first cases were diagnosed.
I’ll continue to update the COVID-19 case and death numbers as long as the global pandemic continues, but as life begins to settle into the new normal in the U.S., I am settling into the new normal of posting a longer, more substantive article every Monday, Wednesday, and Friday.
I would love nothing more than to go back to posting about Australian rules football, but reality seems to be conspiring to keep COVID-19 and racial justice in the news. So today, we examine another way of visualizing COVID-19 cases, one that gives a clearer picture of how the disease is moving around the world.
COVID-19 worldwide statistics
Before I launch into the new visualizations, let’s look at the statistics we have been charting since the beginning of the pandemic, and the predictions we have been tracking since May 17th.
Total cases of COVID-19 diagnosed worldwide: 6,958,337
Total deaths: 401,487
Today is the day we hit 400,000 deaths worldwide, four days after my prediction. The next major milestone is 7,000,000 cases worldwide. I had predicted we would reach that milestone on June 17th, but it looks near certain that we will actually reach seven million tomorrow, a full eight days early.
Deaths per million people by country
The graphs that I have been showing from the beginning have shown the total number of cases and deaths, either worldwide or in specific countries. As I have mentioned before, this is definitely the right way to look at deaths, because once someone dies of COVID-19, they’re not coming back to life. Each new death just adds to the COVID-19 death toll.
So I’ll break from my usual order and show deaths before I show cases. Deaths per million people for selected countries shown on a map:
Deaths per million people in various countries in Europe and the Americas (click for a larger version)
and as a graph showing increasing death from mid-February to today:
Graph of deaths per million people for the usual “Big Ten” countries (click for a larger version)
A new look at case rates
In many ways, the steadily-increasing graphs I’ve been making are the right graphs for case rates too. It is absolutely a meaningful question to ask “how many people have been diagnosed with COVID-19 since the beginning of the epidemic, and how does that vary between countries?” Answers can provide extremely useful data for planning nationwide responses.
But in another way, these cumulative case graphs offer a misleading picture of the current state of the pandemic. The dead stay dead, but people who are sick with COVID-19 can get better. Although it’s still not clear whether having COVID-19 confers immunity to future infection, it is both meaningful and encouraging to plot how many people are currently sick. While it’s hard to know who is sick at any moment, it’s quite easy to plot the number (or rate) of new cases diagnosed.
To recap: for each day on my graphs, I have been plotting the total number of cases (overall or per million people) that have been reported as of that day. For today’s graphs, I will instead plot the number of NEW cases reported each day.
IMPORTANT NOTE: These graphs CANNOT be directly compared with the cumulative case graphs that I have been making since March 18th. Please don’t even try to compare them. They show the same information in a completely different way.
One possible downside of this approach is that the number of new cases reported on a particular day is subject to large day-to-day variations for a variety of reasons including statistical dumb luck. These daily variations are quite interesting and will be the subject of a future post. But they make it difficult to see overall trends, so I am “smoothing” the curves by replacing each day’s case data with the average value itself, the 5 days before, and the 5 days after. This results in a curve that has approximately the same value as the real curve everywhere, but with the daily variations removed so the curve looks much smoother.
It’s easier to explain with an example: the graph below shows the number of cases of COVID-19 reported worldwide, every day since outbreak began in Wuhan, China. The blue line shows the actual number of cases reported each day; the red line shows the smoothed curve.
New cases reported each day worldwide: reported data (blue) and smoothed data (red)
In case you think I’m doing something weird to try to mislead you: no I’m not. This is a very common method in all areas of science. Also, look at the curves, they clearly look about the same.
For the rest of the graphs in this post, I will show ONLY the smoothed data. But it’s easy to create graphs using the observed data instead. If you’d like to try, download my Excel template linked below (version 4) and send me an email (jordan.raddick@gmail.com). I’d be glad to show you how.
So how do our usual “Big 10” countries look with this new visualization approach? Confusing. Don’t try to make sense of the 10 overlapping lines in the graph below, just enjoy how confusing it is.
Better graphs below!
Clearly we need a better approach – and fortunately, our better approach will also allow us to consider some of the countries we have looked at previously.
The only thing I’d like you to notice about the overlapping curves on the graph above is their general shape: the curves rise, reach a peak, and fall. The peak can be narrow or wide, high or low, and it can occur at different times. For some countries, the peak has yet to arrive. (Note: You might be tempted to call these “bell curves,” because they are vaguely bell-shaped – but the term “bell curve” refers to a specific shape of curve, which these are not.)
The shapes of these curves allow us to divide countries into four general categories based on how the epidemic has evolved there. Below, I show the curves for the countries in each category, and briefly explain what they mean. Over time, we’ll follow these categories as different countries move in and out of them. All graphs have the same scale except where noted.
Countries where the epidemic was quickly and easily contained
I’m deliberately showing these countries on the same scale as the countries in other categories (the vertical axis going from 0 to 275 cases per million people); it really drives home how well these countries kept cases to a minimum. All three are found in Asia: China, South Korea, and Japan. For this graph only, the horizontal axis starts in late January, otherwise China we would miss nearly all the cases in China.
Plot of cases reported by day (per million people) in China, Japan, and South Korea (click for a larger version)
Whatever these countries did to contain the epidemic, they did it right. COVID-19 could return anywhere, but it seems unlikely in these countries.
Countries where the epidemic is currently under control
In these countries, the curve has risen, peaked, and fallen to near zero – meaning that very few new cases are currently being reported in these countries. They are all in Europe. In the order in which the peak arrived, they are: Italy, Spain, Switzerland, Slovenia, France, and Belgium.
Plot of cases reported by day (per million people) in Italy, Spain, Switzerland, Slovenia, France, and Belgium (click for a larger version)
These countries struggled to contain the epidemic, but finally won – at least for the moment. That’s why I say “currently under control.” Antlers crossed that COVID-19 remains under control in all these countries.
Countries where the epidemic is mostly headed in the right direction
The next category contains countries where the case rate seems to have peaked, and the curve appears to be on a downslope – meaning fewer and fewer cases are being reported each day. These countries are: the United States, the United Kingdom, Russia, Belarus, Peru, and Qatar. This is the only graph with an inset, otherwise Qatar would have been literally off the charts. The vertical axis of the inset graph goes from zero to 700 cases per million.
These countries have taken concrete steps to fight the pandemic, and those steps are working. If they continue on their current course, they will join the list of countries that have the disease under control. The only thing that worries me is that the decrease has stopped in the U.S. It’s not going up, but it’s not going down either. Hopefully it will work its way back down soon.
Countries where the epidemic is getting worse
Last and unfortunately most are those countries where the epidemic continues to get worse: Chile, Saudi Arabia, Brazil, Sweden, and India:
I’ll continue making these kinds of graphs in future updates.
Want to try out some other ways for yourself? You can get the data that I used to make these graphs from the European Centers for Disease Control’s Coronavirus Source Data; choose “all four metrics.” The Excel template that I used to make the plot of deaths is version 3.3. The plots of new cases by day come from version 4. You want the sheet that says “BIG TEN,” starting at column BI. I’d love to see what you can build with it, and I’m happy to help you figure it out! You can reach me by email at jordan.raddick@gmail.com, or leave a comment here.
Update tomorrow, and every day after that until this pandemic comes to an end. And coming Wednesday: finally a break from COVID-19 and racial justice for some imaginary sports.
Graphs day 82, pandemic day 88, day 159 since the first cases were diagnosed.
Total cases of COVID-19 diagnosed worldwide: 6,835,723
Total deaths: 398,636
It looks like we’ll hit 400,000 deaths worldwide tomorrow, four days after my initial prediction. The maps are back today! Also India is back in the graphs, shown by the chocolate brown line.
Cases per million people by country
Cases per million people by country, first with a map of today’s case rates per million people in various countries in Europe and the Americas:
Cases per million people – click for a larger image
and as a graph comparing case rates on an even scale from each country’s “day zero,” the date on which cases reached 1 in 1,000,000:
Cases per million people in our “Big 10” countries – click for a larger image
Deaths per million people by country
Deaths per million people for selected countries shown on a map:
Deaths per million people on a map of Europe and the Americas
and as a graph on an even scale:
Deaths per million people in our “Big 10” countries – click for a larger image
Coming up tomorrow: back on the regular Monday-Wednesday-Friday posting schedule, I’ll look at some other ways of plotting case data.
Want to try out some other ways for yourself? You can get the data that I used to make these graphs from the European Centers for Disease Control’s Coronavirus Source Data; choose “all four metrics.” You are welcome to use my Excel template (version 3.3). I’d love to see what you can build with it, and I’m happy to help you figure it out!
Update tomorrow, and every day after that until this pandemic comes to an end.
Graphs day 81, pandemic day 87, day 158 since the first cases were diagnosed.
Not actually Jeremy, but close enough
Total cases of COVID-19 diagnosed worldwide: 6,706,022
Total deaths: 394,713
In today’s update, we look at how cases and deaths stack up against my predictions, and learn an important lesson about data visualization. And at the request of my awesome friend Jeremy, we look at cases and deaths in Ireland, in place of India. As you’ll see, there’s not much reason to keep following Ireland, so we’ll return to following India tomorrow.
Cases and deaths worldwide
Ever since I issued my predictions on May 17th, I’ve been tracking how my predictions match up with reality. We have continuously seen how the curve of cases (red dots) is running far ahead of the predictions (black line):
When I was graphing deaths on the same scale, it looked like cases were way ahead my prediction, and deaths were slightly behind my prediction. But here’s what happens when I split deaths into a separate graphs (deaths are the purple dots, the prediction is again the black line):
Deaths are falling behind my prediction just as dramatically as cases are rising ahead. There’s an important lesson here:
When you make a graph, always use the most appropriate scale to show the effects you want to show!
Still no maps today, and it’s an odd-numbered day. Let’s look at case and deaths by country from mid-February to the present.
Cases per million people by country
Cases per million people by country – India is gone this week in favor of Ireland (lavender). All countries are labeled as usual; percentages indicate the case fatality rate in each country, from 0.6 percent in Belarus to 11 percent in Sweden. Also note that I fixed the image so you can click on it for a larger version.
The curve for Ireland is high, but it’s been almost completely flat since late May. Meanwhile, Chile is the Energerizer Bunny of countries – cases keep going and going and going…
Deaths per million people by country
Same color scheme as before, but now showing deaths per million people by country – again with Ireland as lavender, and again clickable:
Deaths in Ireland are just as flat as cases. So there’s no reason to continue tracking Ireland. We’ll return to tracking India tomorrow.
Want to try it yourself? Try it yourself! You can get the data that I used to make these graphs from the European Centers for Disease Control’s Coronavirus Source Data; choose “all four metrics.” You are welcome to use my Excel template (version 3.3). I’d love to see what you can build with it, and I’m happy to help you figure it out!
Update tomorrow, and every day after that until this pandemic comes to an end.
