Urban Heat Islands

Yesterday was the first day of 32º+C (90º+F) in Philadelphia in October in 78 years. Gross. But it made me remember this piece last month from NPR that looked at the correlation between extreme urban heat islands and areas of urban poverty. In addition to the narrative—well worth the read—the piece makes use of choropleths for various US cities to explore said relationship.

My neighbourhood's not bad, but thankfully I live next to a park.
My neighbourhood’s not bad, but thankfully I live next to a park.

As graphics go, these are effective. I don’t love the pure gradient from minimum to maximum, however, my bigger point is about the use of the choropleth compared to perhaps a scatter plot. In these graphics that are trying to show a correlation between impoverished districts and extreme heat, I wonder if a more technical scatterplot showing correlation would be effective.

Another approach could be to map the actual strength of the correlation. What if the designers had created a metric or value to capture the average relationship between income and heat. In that case, each neighbourhood could be mapped as how far above or below that value they are. Because here, the user is forced to mentally transpose the one map atop the other, which is not easy.

For those of you from Chicago, that city is rated as weak or no correlation to the moderately correlated Philadelphia.

I lived near the lake for eight years, and that does a great deal for mitigating temperature extremes.
I lived near the lake for eight years, and that does a great deal for mitigating temperature extremes.

Granted, that kind of scatterplot probably requires more explanation, and the user cannot quickly find their local neighbourhood, but the graphics could show the correlation more clearly that way.

Finally, it goes almost without saying that I do not love the red/green colour palette. I would have preferred a more colour-blind friendly red/blue or green/purple. Ultimately though, a clearer top label would obviate the need for any colour differentiation at all. The same colour could be used for each metric since they never directly interact.

Overall this is a strong piece and speaks to an important topic. But the graphics could be a wee bit more effective with just a few tweaks.

Credit for the piece goes to Meg Anderson and Sean McMinn.

Baby You Can Drive My Non-automobile Personal Mode of Transportation

Last week was the climate summit in New York, and the science continues to get worse. Any real substantive progress in fighting climate change will require sacrifices and changes to the way our societies function and are organised, including spatially. Because one area that needs to be addressed is the use of personal automobiles that consume oil and emit, among other things, carbon dioxide. But living without cars is not easy in a society largely designed where they are a necessity.

But over at CityLab, Richard Florida and Charlotta Mellander created an index trying to capture the ability to live without a car. The overall piece is worth a read, but as usual I want to focus on the graphic.

The Northeast is where it's at with its dense cities designed for a pre-automobile era
The Northeast is where it’s at with its dense cities designed for a pre-automobile era

It’s nothing crazy, but it really does shine as a good example of when to use a map. First, I enjoy seeing metro maps of the United States used as choropleths, which is why I’ve made them as part of job at the Philly Fed. CityLab’s map does a good job showing there is a geographic pattern to the location of cities best situated for those trying to live a car-free life. Perhaps not surprisingly, one of the big clusters is the Northeast Corridor, including Philadelphia, which ranks as the 17th best (out of 398) and the 7th best of large metro areas (defined as more than one million people), beating out Chicago, ranked 23rd and 8th, respectively.

Design wise I have two small issues. First, I might quibble with the colour scheme. I’m not sure there is enough differentiation between the pink and light orange. A very light orange could have perhaps been a better choice. Though I am sympathetic to the need to keep that lowest bin separate from the grey.

Secondly, with the legend, because the index is a construct, I might have included some secondary labelling to help the reader understand what the numbers mean. Perhaps an arrow and some text saying something like “Easier car-free living”. Once you have read the text, it makes sense. However, viewing the graphic in isolation might not be as clear as it could be with that labelling.

Credit for the piece goes to David Montgomery.

Hog Wild

So admittedly this post should have been up last week, but I liked the lunar cycle one too  much. But today is Friday and who cares. We made it to the end of the week.

In the wake of the shootings last week, someone on Twitter posed the question:

Legit question for rural Americans – How do I kill the 30-50 feral hogs that run into my yard within 3-5 mins while my small kids play?

And with that the Internet was off. Memes exploded across the social media verse. Thankfully the Washington Post took it seriously and found data on the expanding footprint of hogs in the United States.

Pig problems
Pig problems

The article also points out, however, that the firearm that prompted the discussion, the now infamous AR-15, would also be a poor choice against feral hogs as its too small a calibre to effectively deal with the animals.

Credit for the piece goes to the US Department of Agriculture.

Quantifying Part of the Opioid Crisis

Two weeks ago the Washington Post published a fascinating article detailing the prescription painkiller market in the United States. The Drug Enforcement Administration made the database available to the public and the Post created graphics to explore the top-line data. But the Post then went further and provided a tool allowing users to explore the data for their own home counties.

The top line data visualisation is what you would expect: choropleth maps showing the prescription and death rates. This article is a great example of when maps tell stories. Here you can clearly see that the heaviest hit areas of the crisis were Appalachia. Though that is not to say other states were not ravaged by the crisis.

There are some clear geographic patterns to see here
There are some clear geographic patterns to see here

For me, however, the true gem in this piece is the tool allowing you the user to find information on your county. Because the data is granular down to county-level information on things like pill shipments from manufacturer to distributor, we can see which pharmacies were receiving the most pills. And, crucially, which manufacturers were flooding the markets. For this screenshot I looked at Philadelphia, though I only moved here in 2016, well after the date range for this data set.

It could be worse
It could be worse

You can clearly see, however, the designers chose simple bar charts to show the top-five. I don’t know if the exact numbers are helpful next to the bars. Visually, it becomes a quick mess of greys, blacks, and burgundies. A quieter approach may have allowed the bars to really shine while leaving the numbers, seemingly down to the tens, for tables. I also cannot figure out why, typographically, the pharmacies are listed in all capitals.

But the because I lived in Chicago for most of the crisis, here is the screenshot for Cook County. Of course, for those not from Chicago, it should be pointed out that Chicago is only a portion of Cook County, there are other small towns there. And some of Chicago is within DuPage County. But, still, this is pretty close.

Better numbers than Philly
Better numbers than Philly

In an unrelated note, the bar charts here do a nice job of showing the market concentration or market power of particular companies. Compare the dominance of Walgreens as a distributor in Cook County compared to McKesson in Philadelphia. Though that same chart also shows how corporate structures can obscure information. I was never far from a big Walgreens sign in Chicago, but I have never seen a McKesson Corporation logo flying outside a pharmacy here in Philadelphia.

Lastly, the neat thing about this tool is that the user can opt to download an image of the top-five chart. I am not sure how useful that bit is. But as a designer, I do like having that functionality available. This is for Pennsylvania as a whole.

For Pennsylvania, state-wide
For Pennsylvania, state-wide

Credit for the piece goes to Armand Emamdjomeh, Kevin Schaul, Jake Crump and Chris Alcantara.

The Ebola Outbreak in the Congo

Ebola, which killed 11,000 people in West Africa in 2014 (which I covered in a couple of different posts), is back and this time ravaging the Congo region, specifically the Democratic Republic of the Congo (DRC). The BBC published an article looking at the outbreak, which at 1,400 deaths is still far short of the West Africa outbreak, but is still very significant.

That's looking like a tenuous border right now…
That’s looking like a tenuous border right now…

The piece uses a small multiples of choropleths for western Congo. The map is effective, using white as the background for the no case districts. However, I wonder, would be more telling if it were cases per month? That would allow the user to see to where the outbreak is spreading as well as getting a sense of if the outbreak is accelerating or decelerating.

The rest of the article features four other graphics. One is a line chart that also looks at cumulative cases and deaths. And again, that makes it more difficult to see if the outbreak is slowing or speeding up. Another is how the virus works and then two are about dealing with the virus in terms of suits and the containment camps. But those are graphics the BBC has previously produced, one of which is in the above links.

Credit for the piece goes to the BBC graphics department.

A Shrinking Illinois

Last week we looked at the data on Pennsylvania from the US Census Bureau and found the Commonwealth’s population is shifting from west of the Appalachians to the southeast of the state. That got me thinking about Illinois, one of three states to have experienced a decline in population. Is there a similar geographic pattern evident in that state’s data? (Plus, I lived there for eight years, so I am curious how the state evolved over a similar time frame.)

A lot more red in this map…
A lot more red in this map…

Well, it turns out the pattern is not so self-evident in Illinois as it is in Pennsylvania. Instead, we see small clusters of light blues across a sea of red. In other words, the population decline is widespread, though not necessarily extreme. However, it is notable that in the far south of the state, Alexander County, home to the city of Cairo, has seen the greatest decline in population since 2010, not just in Illinois, but in the entire United States (in percentage terms).

Unlike Pennsylvania, where the state’s primary city of Philadelphia is growing (albeit slowly), in Illinois the primary city of Chicago has seen its population shrink over the last several years. However, the counties south and west of Cook County have grown. Kendall County, where parts of Aurora and Joliet are located along with growing towns like Oswego and Plano, grew at over 11%.

The state’s other growing counties fall across the state from north to south, east to west. In the south the county containing the eastern suburbs of Carbondale has grown modestly. But for real percentage growth, one should look west towards Monroe County, a southern suburb of St. Louis, Missouri located just across the Mississippi River.

Then in the centre of the state we see growth in McLean and Champaign Counties. The former is home to Bloomington and Normal. While Champaign is home to the eponymous city as well as its neighbour, Urbana.

All in all, the pattern that emerges is that of urban/suburban vs. rural. With some notable exceptions, e.g. Cook County, the only growth in Illinois is in counties that have prominent cities or towns. Meanwhile, rural counties shrink—the aforementioned Alexander most notably.

Credit for this piece is mine.

Pennsylvania’s Population Shifts

Last month the US Census Bureau published their first batch of 2018 population estimates for states and counties. Pennsylvania is one of those states that is growing, but rather slowly. It will likely lose out to southern and western states in the 2020 census after which House seats will be reapportioned and electoral college votes subtracted.

From 2018 to 2010, the Commonwealth has grown 0.8%. Like I said, not a whole lot. But unlike some states (Illinois), it is at least growing. But Pennsylvania is a very diverse state. It has very rural agricultural communities and then also one of the densest and largest cities in the entire country with the whole lot in between . Where is the growth happening—or not—throughout the state? Fortunately we have county-level data to look at and here we go.

Some definite geographic patterns here…
Some definite geographic patterns here…

The most immediate takeaway is that the bulk of the growth is clearly happening in the southeastern part of the state, that is, broadly along the Keystone Corridor, the Amtrak line linking Harrisburg and Philadelphia. It’s also happening up north of Philadelphia into the exurbs and satellite cities.

We see two growth outliers. The one in the centre of the state is Centre County, home to the main campus of Pennsylvania State University. And then we have Butler County in the west, just north of Pittsburgh.

The lightest of reds are the lowest declines, in percentage terms. And those seem to be clustered around Scranton and Pittsburgh, along with the counties surrounding Centre County.

Everywhere else in the state is shrinking and by not insignificant amounts. Of course this data does not say where people are moving to from these counties. Nor does it say why. But come 2020, if the pattern holds, the state will need to take a look at its future planning. (Regional transit spending, I’m looking at you.)

Trump-won Counties Are Winning

Yesterday we looked at how China and the European Union are planning their tariff/trade war retaliation to target Trump voters. Today let’s take a look at how those voters are doing as this article from Bloom does.

Lots of green, but some noticeably red counties in Florida.
Lots of green, but some noticeably red counties in Florida.

The article is not terribly complicated. We have four choropleth maps at the county level. Two of the maps isolate Trump-won counties and the other two are Clinton-won. For each candidate we have a GDP growth and an employment growth map.

In the Trump-won maps, the Clinton-won counties are white, and vice versa. Naturally, because the Democratic vote is greatest in the large cities, which, especially on the East Coast, are in tiny counties, you see a lot less colour in the Clinton maps.

Not a whole lot to see here…
Not a whole lot to see here…

Design wise, I should point out the obvious that green-to-red maps are not usually ideal. But the designers did a nice job of tweaking these specific colours so that when tested, these burnt oranges and green-blues do provide contrast.

Here they appear more of a yellow to grey
Here they appear more of a yellow to grey

But I am really curious to see this data plotted out in a scatter plot. Of course the big counties in the desert southwest are noticeable. But what about Philadelphia County? Cook County? Kings County? A scatter plot would make them equally tiny dots. Well, hopefully not tiny. But then when you compare GDP growth and employment growth and benchmark them against the US average, we might see some interesting patterns emerge that are otherwise masked behind the hugeness of western counties.

But lastly. And always. Where. Are .Alaska. And. Hawaii? (Of course the hugeness problem is of a different scale in Hawaii. Their county equivalents are larger than states combined.)

Credit for the piece goes to the Bloomberg graphics department.

Trade War Retaliation

About a week and a half ago the Economist published an article about the retaliatory actions of the European Union and China against the tariffs imposed by the Trump administration. Of course last week we had a theme of sorts with lineages and ancestry. So this week, back to the fun stuff.

What makes today’s piece particularly relevant is that over the weekend, Trump announced he might increase the tariffs proposed, but not yet implemented, upon Chinese goods. So some economists looked at the retaliatory tariffs proposed by the EU and China.

Ultimately Trump's tariffs are not paid by foreign governments, but by US citizens.
Ultimately Trump’s tariffs are not paid by foreign governments, but by US citizens.

Each targets Trump voters, albeit of different types. But China appears more willing to engage in a brutal fight. Its tariff proposal would not just harm Trump voters, but would also harm Chinese citizens. The EU’s plan appears tailored to maximise the pain on Trump voters, but minimise that felt by its own citizens.

A few minor points. I like how the designers chose to highlight high impact categories with colour. Lower impact shares are two shades of light grey. But after that, the scale changes. I wonder how the maps would compare if each had been set to the same scale. It looks doable as the bottom range of the maximum bin is 6% for the EU and 8% for China. (Their high limit is much higher at 22% compared to the EU’s 10%.)

That said, it does a good job of showing the different geographic footprints of the two retaliatory tariff packages. Tomorrow—barring breaking news—we will look at why that is important.

Credit for the piece goes to the Economist Data Team.

The Great Migration Map

Yesterday in a post about Angela’s forced journey from Africa to Jamestown I mentioned that the Pilgrims arrived at Plymouth Bay just one year later in 1620. From 1620 until 1640 approximately 20,000 people left England and other centres like Leiden in the Netherlands for New England. Unlike places like Jamestown that were founded primarily for economic reasons, New England was settled for religious reasons. Consequently, whereas colonies in Virginia drew young men looking to make it rich—along with slaves to help them—New England saw entire families moving and transplanting parts of towns and England into Massachusetts, Rhode Island, Connecticut, and New Hampshire.

New England kept fantastic records and we know thousands of people. But we do not know whence everyone arrived, but we do know a few thousand. And this mapping project from American Ancestors attempts to capture that information at the English parish level. At its broadest level it is a county-level choropleth that shows, for those for whom we have the information, the majority of the migration, called the Great Migration, came from eastern England, with a few from the southwest.

Quite a few from Norfolk, Suffolk, and Essex
Quite a few from Norfolk, Suffolk, and Essex

You can also search for specific people, in which case it brings into focus the county and the parishes within that have more detail. In this case I searched for my ancestor Matthew Allyn, who was one of the founders of Hartford, Connecticut. He came from Braunton in Devon and consequently appears as one of the two people connected to that parish.

Devon did not have nearly as many people emigrate as the eastern counties
Devon did not have nearly as many people emigrate as the eastern counties
But was Thomas related to Matthew? We don't know.
But was Thomas related to Matthew? We don’t know.

Overall, it’s a nice way of combining data visualisation and my interest/hobby of genealogy. The map uses the historical boundaries of parishes prior to 1851, which is important given how boundaries are likely to change over the centuries.

This will be a nice tool for those interested in genealogy and that have ancestors that can be traced back to England. I might be biased, but I really like it.

Credit for the piece goes to Robert Charles Anderson, Giovanni Flammia, Peter H. Van Demark.