Hot and Not So Hot Graphics

Thankfully today’s forecast calls for cooler temperatures. Your author is not a fan of hot weather, which means being outside in summer is…less than ideal. It also means that the air conditioner runs frequently and on high for a few months. (Conversely, I can probably count on one hand the number of times I turned on the heat this winter.)

The problem is, the two biggest contributors to US carbon emissions? Heating/cooling and transport. In other words, heating your home in the winter, cooling it in the summer, and then driving your non-electric vehicle.

After the recent heatwave in New England, the Boston Globe examined the impact of the heatwave on the environment. The article led with the claim it used four charts to do so. I quibble with that distinction because this is a screenshot of the second graphic.

Illustrating is hard

I mean, it’s not prose text. Rather, we have three factettes paired with illustrations. At the top of this post, I mentioned the impact of transport for a reason. In an ideal world, in order to get carbon emissions under control one of the changes we would need to see is getting people out of their personal automobiles and into mass transit. Subways and light rail are far cleaner and can actually be cheaper for households than car ownership. And so we should be encouraging their use and building more of them.

Look above and you’ll see an icon of a subway car. Except it’s not. The graphic/factette is actually talking about rail cars full of coal that transport fuel from mine to generating station. Those look more like this, from James St. James via Wikimedia Commons.

Small, subtle details matter. And so I’d propose a new icon that tries to capture the industrial coal train, ideally something that I spent more than five minutes on.

Illustrating is still hard

But it breaks the linkage between passenger train and coal train, which is not ideal for the purposes of an article highlighting the environmental impacts of US households.

That all said, the article did a really good job with the other graphics it used. My favourite was this chart, decidedly not a combination chart.

Making charts can also be hard, though

It looks at the correlation between high temperatures and energy usage. But, instead of lazily throwing the temperatures atop the bars, the designers more carefully placed them below the energy usage chart. The top chart should look familiar to those who have been following my Covid-19 charts, a daily number that then has the rolling seven-day average plotted above it to smooth out any one-day quirks. The designer then chose to highlight the heatwave in red.

For temperatures, I like the overall approach. But I wonder if a more nuanced approach could have taken the graph a step farther to excellent. Presently we have a single red line representing daily average high temperature. But in the plot above we use red to indicate the heat wave of early June, five consecutive days of temperatures in excess of 90ºF. What if that line were black or grey or some neutral colour, and then only the heatwave was coloured in red? It would more clearly link the two together. And it avoids the trap of red implying heat, when you need to only go back to late May when the East Coast had early spring like temperatures near 50ºF, decidedly not red on a temperature scale.

Overall, though, it’s refreshing to see a thoughtful approach taken here instead of the usual slapdash throw one chart atop the other.

And the rest of the article uses restrained, smart graphics as well. Bar charts and small multiples to capture air pollution and EMS calls. You should read the full article for the insights and the feedback loops we have.

After all, it’s not that the heating/cooling is itself the problem, especially since the removal of CFCs since the Montreal Protocol in 1987 that banned those pesky chemicals that harm the ozone layer—remember when that was the big environmental issue in the 1990s? The issue is how we generate the electricity that powers the heating/cooling systems—and if you want to use electric cars, whence comes their electric charge—as if we’re using coal plants, that just exacerbates the problem. But if we use carbon-less plants, e.g. nuclear, solar, or wind, we’re not generating carbon emissions.

Credit for the piece goes to John Hancock.

Back to the Office, Back to Basics

Two weeks ago I posted about an article from the BBC that used graphics about which I was less than thrilled. Inconsistent use of axis lines, centring the graphic were two of the things that irked me. Two weeks hence, I do want to draw some positive attention to another article in the BBC. This one discusses the, for many of us, impending return to the office. (I’ve also heard the phrase “return to work”, although a coworker of mine pointed out that’s not a great phrase because many of us never stopped working when we decamped for our flats and houses.)

The article discusses why some think the return to a five-day office week will occur within the next few years. There is some sound logic to the idea and for those like your author who are closely following the issue, I recommend the article.

But that’s not why we’re here, instead I wanted to focus on the one data visualisation graphic in the piece. It displays the amount of office space used in the city centres of six different UK cities outside London.

But what about Slough?

Here we have small multiples with the same fixed y-axis display. Axis lines are present and consistent and the baseline is distinct from the other lines. Solid improvement over what we discussed two weeks ago.

My only quibble? The colours here are not necessary. A single colour would work because each city’s graphic exists apart from the rest. The charts also all represent the same type of data, occupied office space. If the chart were doubling or tripling up cities somehow—though I wouldn’t want to see this as a stacked area chart—I would buy the need for colours to differentiate the cities. This, however, represents an opportunity to use a single, BBC-branded colour to define the experience whilst not negatively impacting the communication from the data visualisation standpoint.

Again, though, that’s a minor quibble. Of course, the BBC puts out copious amounts of content daily and I see only a fraction, but it is nice to see an improvement. Furthermore, at the end of the article I also spotted a graphic credit, which I don’t often see—and honestly cannot recall when I last saw period—from the BBC.

I wonder if moving forward the BBC intends to highlight the contributors to articles who are not solely the writers, i.e. the people creating the graphics? Of course, if we did that, we should also probably take a look at the copy editors who also play a role. Especially for an online article as opposed to say a print newspaper or magazine where space is money.

Credit for the piece goes to Daniele Palumbo.

The Month That Lasted a Year

Two Fridays ago I received my second dose of the vaccine. In other words, I’m fully vaccinated and can resume doing…things. Anything. And so this piece from xkcd seemed an appropriate way to wrap up what has been a horrible, no good, terrible year.

The longest month of our lives.

Credit for the piece goes to Randall Munroe.

Covid-19 Update: 13 September

Apologies for the lack of posting last week. I’m on deadline for, well, today. Plus I had some technical difficulties on the server side of the blog. But it’s a Monday, so we’re back with Covid updates for Pennsylvania, New Jersey, Delaware, Virginia, and Illinois.

New cases curves for PA, NJ, DE, VA, & IL

The good news, such that it is during a global pandemic, is that in Pennsylvania, Delaware, and Illinois, the seven-day average appears to be lower than this time last week or, especially in Delaware’s situation, about to break. For the First State, I’m looking at those days prior to the weekend below the average line that, in combination with the weekend, will likely begin to push that trend downward, especially if we keep seeing fewer and fewer cases this week.

Unfortunately, some states like Virginia and New Jersey appear to be, not surging, but experiencing low and slow growth. Low and slow, while great for barbecue, is less than ideal during a pandemic. Granted, it’s better than the rapid infections we saw in March, April, and May, but it still means the virus is spreading in those communities.

Death curves in PA, NJ, DE, VA, & IL.

When we look at deaths from Covid-19 in these five states, the news is better. The only real significant level of deaths was in Virginia, but we can see that the latest little surge, which was at peak last week, has now all but abated, almost to a level not seen since the spring.

The other states remain low with, at most, deaths average about 20 per day. Again, not good, but better than hundreds per day.

Credit for the pieces is mine.

Covid-19 Update: 7 September

Yesterday was a holiday in the States, and so let’s begin this shortened week with a look at the Covid situation in Pennsylvania, New Jersey, Delaware, Virginia, and Illinois.

If we compare this morning’s charts of yesterday’s data to last Monday’s, we can see some concerning trends.

New case curves in PA, NJ, DE, VA, & IL.

In Pennsylvania, that includes a rising trend. Anecdotally, that might be tied to the outbreaks in and around universities. We see rising trends in Delaware and Virginia as well, though some of Delaware’s new numbers might be tied to some cases that failed to initially make into the state’s digital database. And so as the state begins to enter them now, it artificially inflates the new case numbers.

Illinois had an enormous spike of cases from a backlog that the state entered, over 5,000 new case in that one day. That’s going to mess with the average trend given the size of the anomaly. So we’ll need to wait until later this week to see where the trend really is.

Then in terms of deaths, the most worrying state was Virginia which last week was mid-peak. But that appears to maybe be trending back down. Though the data we have does include two day’s of weekend numbers and Tuesday’s numbers, instead of the usual “rebound” will be more of the usual weekend depressed numbers.

Death curves in PA, NJ, DE, VA, & IL.

Credit for the pieces is mine.

Covid-19 Update: 30 August

After dealing with hurricane forecast plots last Monday, we’re back to the nature-made, man-intensified disaster of Covid-19 in the United States. So in the five states we review, where are we with the pandemic?

Compared to the charts from two weeks, looking at daily new cases, in some places we are in a better spot, and in others not much has changed. In fact Illinois is the only place worse off with its seven-day average higher than it was two weeks ago, but not by dramatically much.

New cases curve in PA, NJ, DE, VA, & IL.

In fact we see in Pennsylvania, New Jersey, and Delaware that the average number of daily new cases is lower than it was two weeks ago. Virginia dipped lower, but has recently returned to approximately the same level and in that sense is in no different a place. Of course the key factor is how those trends all change over the coming week.

But what about in terms of deaths?

Deaths curve in PA, NJ, DE, VA, & IL.

Well here there is bad news in Virginia. Two weeks ago a spike in deaths there had largely subsided. Two weeks hence? We are in the middle of a third spike of deaths, reaching nearly 20 deaths per day.

Fortunately, the other four states remain largely the same, and that means few deaths per day. Indeed, for Pennsylvania and New Jersey that means deaths in the low double-digits or often in the single digits. Delaware has not reported a new death in four days. And Illinois, while up a little bit, is in the low single-digits, but generally just a few more deaths per day than Pennsylvania and New Jersey.

Credit for the piece is mine.

Covid-19 Update: 16 August

So here are the charts from the last week of Covid data in Pennsylvania, New Jersey, Delaware, Virginia, and Illinois.

When we compare last week’s update to today’s, we can see that Pennsylvania did indeed bottom out and is back on the rise and the same can probably be said for Delaware. Although a fair amount of the one-day spikes in those numbers we see today are from an outbreak in a correctional system.

Whilst Virginia did go up, by week’s end, it had settled back down to a point not dissimilar to last week. So nothing really changed and time stood still in Virginia. The same can also be loosely said for New Jersey, where it was more about fluctuations than determined rises or falls.

In Illinois, however, we finally saw a plateauing of the new cases numbers and with the slightest of declines .

New cases in PA, NJ, DE, VA, and IL.
New cases curve in PA, NJ, DE, VA, and IL.

Then in deaths we have not much to say as they remain low in New Jersey and Delaware and stable and moderate in Illinois.

Virginia’s recent spike appears to have subsided, as it’s back to nearly 10 deaths per day from the virus.

But most concerning is Pennsylvania. Here, while the numbers are still relatively low, they are on a slow and gradual rise. At this point the seven-day average is beginning to rise above 20 deaths per day.

Deaths in PA, NJ, DE, VA, and IL
Death curve in PA, NJ, DE, VA, and IL

Credit for the piece is mine.

When Is an Opener Game in Baseball Really Just a Bullpen Game?

Whenever someone not named Eovaldi or Perez starts a Red Sox game in 2020, that’s when.

We all know the Red Sox are the worst team in the American League. They have only two starters, maybe sometimes a third. And then the last two days of the regular five-day rotation cycle, manager Ron Roenicke throws some relievers at the wall and sees which ones stick that night. Spoiler: Few do.

But as much as I enjoy listening to the three-man broadcasting booth (Remy and Eck make the games fun to at least listen to) the games are unwatchable. And then to hear them try and dress a game up as having an opener? Well, what is the opener?

For the non-baseball fans, most are probably aware enough that some guy goes out to a small hill and throws (pitches) a ball at a batter for most of the night. Then towards the end, when the guy’s energy wanes, he is replaced by some guy who throws really fast. That’s over simplified, but that’s a normal ballgame. A starting pitcher records five, but ideally at least six, innings of work before handing the ball over to an eighth-inning setup man and then a ninth-inning closer. Sometimes a really good seventh-inning reliever sets up the setup man.

A bullpen game, by contrast, is when a bunch of those relief pitchers handle the entire game. Usually this would be after a game went into extra innings (since baseball cannot end in a tie, unless you’re in an All Star game), and the next day’s starting pitcher had to finish the long game by pitching several innings. With nobody available to throw six innings, a bunch of relievers come in and try to cover that by pitching one, two, or three innings each.

The opener game is relatively new. The idea is in addition to the really good closer, a really good opener records the first inning or two (3–6 outs) to deal with the opposing team’s best hitters. He then hands the ball over to a mediocre starting pitcher who throws the next four or five innings, who then hands the ball over to the late-inning relief specialists. Doing it this way, the starter avoids one set of at-bats or plate appearances by the opposition’s best hitters.

But when is an opener just a bullpen game? Well, it’s when that mediocre starting pitcher isn’t really a starting pitcher. And when he doesn’t even throw four or five innings. Basically all the Red Sox games this year.

I made a graphic this morning to contrast those different types of games and compare them to a game I watched two nights ago between the Red Sox and the Tampa Bay Rays. The game was teed up as an opener with a good relief pitcher by the name of Ryan Brasier starting the first inning. But then instead of a mediocre starter pitching four or five innings, we got a mediocre reliever pitching three innings. He handed it over to a guy who was supposed to go maybe two, but couldn’t get through his second inning. He handed it over to another guy, who handed it over to another guy, who handed it over to a final guy. And none of those last guys were the good relievers you would typically expect to see. (Though, to be fair, the Sox weren’t winning, so why use your best relievers?)

What is an opener and how does it compare to a bullpen game?
Different game types compared

Credit for the piece is mine.

Credit for the Red Sox dumpster fire of a season goes to John Henry and ownership.

Covid-19 Update: 9 August

Weekend data means, usually, lower numbers than weekdays. And with the exception of Delaware that’s what we have today. Some drops, like Illinois, are more dramatic than others, like New Jersey. And so we look at the seven-day trend.

And that tells a slightly different story. On the one hand we have states like Virginia and Illinois that appear to be continuing upward. The rise in Illinois has been slow and steady, but the average is approaching nearly 2000 new cases per day. In Virgina, the rise was more abrupt and the question is whether this peak has crested in recent days or if come the middle of next week it will resume rising.

In New Jersey and Delaware we see two states with does declines after some sudden spurts of new cases. Jersey had risen to nearly 500 new cases less than two weeks ago, but that’s now back down to fewer than 350. And in Delaware, while today’s number is greater than yesterday’s, the trend is still downard after being at over 100 new cases per day two weeks ago.

New cases curves for Pennsylvania, New Jersey, Delaware, Virginia, and Illinois.
New cases curves for Pennsylvania, New Jersey, Delaware, Virginia, and Illinois.

Then we have Pennsylvania. At one point doing it had done so well in controlling the outbreak to bend the curve to fewer than 500 new cases per day at one point. Then as the state began to reopen, cases began to rise again in the west and now the east. But over the last week that statewide average began to fall. But in the last two days that fall appears to have potentially bottomed out. So come the middle of next week, the question will be does the downward trend continue or has the state hit a new valley before another rise?

Finally, in terms of new deaths, with the exception of Virgina, we have yet to see any rise in deaths that might correlate with the recent rises in new cases. And so nothing new there. But it’s worth pointing out that New Jersey has now reached the high single digits in terms of daily deaths from Covid-19. That’s remarkable for a state that back in April saw nearly 300 people dying every single day.

New death curves for Pennsylvania, New Jersey, Delaware, Virginia, and Illinois.
New death curves for Pennsylvania, New Jersey, Delaware, Virginia, and Illinois.

Credit for the graphics is mine.

Covid-19 Update

As I mentioned last week, I am going to try using my blog here for the weekly update on the five states people have asked me to explore. And for the second week in a row, we are basically seeing numbers down compared to previous days. But given that numbers are generally lower on the weekends, that is not terribly surprising.

The real question is by Friday, will these numbers have rebounded?

The Covid-19 curves for PA, NJ, DE, VA, and IL
The Covid-19 death curves for PA, NJ, DE, VA, and IL.

Credit for these graphics is mine.