Lightnews — Scholar-powered news

Juuso Koponen

@juusokoponen.bsky.social

1.5K followers 1.9K following 960 posts

📊 Information designer/data journalist, Koponen+Hildén Co-author, “Data Visualization Handbook” Teacher, data journalism & visualization at Haaga-Helia etc. DJ ❤️ Maurizio, XCOM 2, GitS, Vernaccia di S. Gimignano Output in 🇫🇮🇬🇧 Input also in 🇸🇪🇮🇹

Posts Media Videos Starter Packs

Juuso Koponen @juusokoponen.bsky.social · 12h

But I stand by my position that there are many, many, different use cases for data visualizations, and different solutions work in those different cases. Though area is not the best visual variable, it is sometimes among the good ones for a particular use case.

Juuso Koponen @juusokoponen.bsky.social · 12h

Obvs the grid cartogram on the right works better – I have used this exact example in my (and @jhilden.bsky.social’s) book and in many lectures – but it is exceptionally good, and it also gives a reasonably good amount of geographic context. The type of thing I believe readers hate is more like this

A cartogram by The Economist showing the vote share of nationalist parties

1 2

Juuso Koponen @juusokoponen.bsky.social · 16h

I tend to agree, but would argue that the optimal shape may depend on the geographic pattern. Circles are an all-around safe bet, but patterns where the observations concentrate on a narrow strips, such as along rivers, roads, tectonic plate boundaries, etc. can work better with bars/triangles.

A map of earthquake magnitudes in 2023, using pointed triangles to depict the magnitude.

1 1

Juuso Koponen @juusokoponen.bsky.social · 16h

Of course, if the task was to compare annual power output at each facility one-to-one, the map would not work. For that task, we should pick another type of visualization. But for a bunch of other tasks the map is superb, by far.

Juuso Koponen @juusokoponen.bsky.social · 16h

This map, although using a less-optimal visual variable for encoding quantities, outperforms cartograms, tables, and bar charts for the task at hand: understanding where and by what means Switzerland produces its electricity (or, produced in 1971).

Juuso Koponen @juusokoponen.bsky.social · 16h

Take this map from Imhof’s “Atlas der Schweiz,” showing power plants by type and yearly production. Here areas work perfectly well: We see that nuclear plants are huge in terms of output, oil-fired plants and hydropower vary more. We see many of the largest plants cluster near northern border. Etc.

A map showing yearly electricity production of each Swiss power plant as circles of varying sizes and colors. The size corresponds to the annual electricity production, the color to the type of plant. Hydropower plants also have more detail about the exact type of plant as shapes within the circle.

1 3

Juuso Koponen @juusokoponen.bsky.social · 16h

Disagree. There are so many very, very different use cases for maps and other visualizations that such hard and fast rules rarely make sense. (Also, my experience is that readers *hate* equal-area cartograms.)

2 1

Juuso Koponen @juusokoponen.bsky.social · 21h

Well, there are use cases where areas tend to work reasonably well compared to the alternatives. For example, for showing absolute quantities on a map.

Juuso Koponen @juusokoponen.bsky.social · 23h

More comedy shows should have Powerpoint presentations and infographics!

1 1

Juuso Koponen @juusokoponen.bsky.social · 1d

Correction: yes & no votes were tied in the S&D caucus. But even that much agreement with the right on this issue is, well, absurd.

Juuso Koponen @juusokoponen.bsky.social · 1d

For precise comparisons, this is, of course, the case. But I would argue area (as a visual variable) is not well suited for that task, in any case. For the more common tasks of perceiving broad size categories, clustering, etc., Flannery works without explanation, and, I think, slightly better.

Juuso Koponen @juusokoponen.bsky.social · 1d

I also use RAWGraphs, when there are many circles to scale. For just a handful, I do the calculation by hand. But I’ve found it is a smaller step for graphic design students to take to learn to calculate this manually than to move to using RAWGraphs… 😅

1 1

Juuso Koponen @juusokoponen.bsky.social · 1d

I had to delete the post you were replying to because there was a weird mistake in the graphic. I reposted it with the corrected graphic, but obviously you reply is no longer connected to it. Sorry! 😅

Juuso Koponen @juusokoponen.bsky.social · 1d

For anyone interested, here is another way of doing the calculation, using both the square root method and the Flannery method. (The only thing that changes between them is the exponent.)

A graphic titled “Calculating symbol sizes”
Underneath it, two formulas:
Square root method: D equals brace open V divided by V ref brace closed to the power of 0.5 times D ref
Flannery method: D equals brace open V divided by V ref brace closed to the power of 0.57 times D ref
(The only difference between the two formulas is the exponent, either 0.5 or 0.57.)
The terms are explained thus:
D = diameter of the symbol at hand
V = value of the data point at hand
D ref = diameter of the reference symbol
V ref = value of the reference data point

Below there is a visual demonstration of how different values should be converted into circles of different diameters in comparison to a reference circle that has the value of 1 million and diameter of 300 pixels.

1 3

Juuso Koponen @juusokoponen.bsky.social · 1d

In cartography, it is relatively common to use the exponent 0.57 instead of 0.5 when scaling areas, as suggested by James Flannery based on his empirical research on perceived areal differences. This does not, of course, solve the problem entirely, but may help a bit.

1 2

Juuso Koponen @juusokoponen.bsky.social · 1d

Agreed. I do think area is useful for depicting secondary, less important dimensions in the data, when position is used to encode the main dimension(s). In a map, for example, the geographic position is obviously encoded using position, so areas can be used to show magnitudes on top of that.

2 4

Juuso Koponen @juusokoponen.bsky.social · 1d

There are many ways to calculate the square root. Calculators usually have a button for it. In Excel, you can use the =SQRT() function. Also, because square root is the same thing as the “0.5th” power, you can also calculate it with a formula such as =10^0.5 (for the square root of 10). 2/2

Juuso Koponen @juusokoponen.bsky.social · 1d

📊 #Dataviz PSA: When using areas to depict numbers, make sure you are indeed scaling the *area*, and not the width/length/diameter of the circles (or other shapes). Because software such as Illustrator don’t let you specify the area directly, you need do a small calculation to get it right. 1/2

A graphic with the following text: “The correct way to scale up or down dimensions when using areas to show values: use a scale factor to multiply the square root of the value.” Below the text is an equation that reads: width equals square root of value times scale factor. Further below are two circles of varying sizes. One is labeled: ”Value: 10, Area: 1” and below it a line segment, corresponding to the circle’s diameter is labeled ”3.16 times 3.16 equals 10 millimeters”. The other is labeled: ”Value: 50, Area: 5 times 1” and the line segment below it is labeled ”7.07 times 3.16 equals 22.34 millimeters”.

4 9 25

Juuso Koponen @juusokoponen.bsky.social · 1d

It’s not just the right, though. The majority of S&D MEPs also voted for this.

1 2

Reposted by Juuso Koponen

Attila Bátorfy @attilabatorfy.bsky.social · 3d

Swedish people in front of charts
Source: digitaltmuseum.org

1 6 13

Juuso Koponen @juusokoponen.bsky.social · 3d

BKT:n käyttäminen kaikkein keskeisimpänä talousindikaattorina – eli tapa jolla toimimme nyt – on kuin arvioisi yrityksen menestystä vain liikevaihdon kasvun perusteella, huomioimatta lainkaan toiminnasta syntyvää voittoa tai tappiota ja taseen kehitystä.

Juuso Koponen @juusokoponen.bsky.social · 3d

Kiinnostavaa tietoa kansantalouden tilinpidon uudistuksesta 2030. Pääoman ja luonnonvarojen kertymisen ja kulumisen huomioiva nettokansantuote olisi (vihdoin!) nousemassa keskeiseksi indikaattoriksi. Sen kasvun nostaminen tavoitteeksi BKT:n sijaan veisi päätöksentekoa selvästi järkevämpään suuntaan.

Tilastokeskus @stat.fi · 3d

Bruttokansantuotteen laskentatapa on muuttumassa, vaikka tilastotuotannon kansainvälisen kehityksen rattaat pyörivät rauhallisesti. Keväällä hyväksytty suositus tulee huomioimaan paitsi luonnonvarat, myös hyvinvoinnin, kirjoittavat asiantuntijamme Tieto&trendit-artikkelissa:
stat.fi/tietotrendit...

Kuvassa on teksti “Miten tulot, kulutus ja varallisuus jakautuvat?”. Lisäksi oikealla näkyy valokuvassa rakenteilla olevia kerrostaloja ja torninostureita.

1 5

Reposted by Juuso Koponen

Atte Harjanne @atteharjanne.vihreat.fi · 4d

Uusiutuvan energian lisääminen on nimenomaan Suomessa ollut yksi perustelu metsäenergian kestämättömälle käytölle. Kaikki uusiutuva ei ole kestävää eikä kaikki kestävä uusiutuvaa (muistakaamme mm. ydinvoima).

2 1 15

Juuso Koponen @juusokoponen.bsky.social · 4d

Tässä on kaikki eläkelajit mukana – myös työkyvyttömyyseläkkeet, jotka ovat yleensä vanhuuseläkkeitä pienempiä. Osatyökyvyttömyyseläke ei oikeuta takuueläkkeeseenkään.

Juuso Koponen @juusokoponen.bsky.social · 4d

A beautiful sight to behold! 😍

2 1