Tag: dataviz

Posted on
by

This is definetely a word I’ll remember: data visceralisation.
The term is suggested for data visualization in virtual reality, so that people can better experience differences in data, understand them viscerally.

It is something that I think definitely is useful, not just in virtual reality but also in making data visualisation physical, which I called ‘tangible infographics’ in 2014. You switch the perspective to one or more other senses, thus changing the phenomenological experience, which can yield new insights.

In both, tangible infographics and data visceralisation, the quest is to let people feel the meaning of certain datasets, so they grasp that meaning in a different way than with the more rational parts of their mind. (Hans Rosling’s toilet paper rolls to convey global population developments come to mind too).

Benjamin Lee et al wrote a paper and released a video exploring a number of design probes. I’m not sure I find the video, uhm, a visceral experience, but the experiments are interesting.

They look at 6 experimental probes:

  1. speed (olympic sprint)
  2. distance (olympic long jump)
  3. height (of buildings)
  4. scale (planets in the solar system)
  5. quantities (Hong Kong protest size)
  6. abstract measures (US debt)

The authors point to something that is also true for the examples of 3d printed statistics I mentioned in my old blog post which are much less useful with ‘large numbers’ because the objects would become unwieldy or lose meaning. There is therefore a difference between the first three examples, which are all at human scale, and the other three which aim to convey something that is (much) bigger than us and our everyday sense of our surroundings. That carries additional hurdles to make them ‘visceral’.

(Found in Nathan Yau’s blog FlowingData)

Posted on
by

Peter, like me getting to grips with Webmention, has now used it to send all his own old postings a webmention where he links to them retroactively. So now in his comment database he has a full list of all the links between his own postings.

He says “I wish I had a way of visualizing the interconnections between my posts“.
This type of thing is of interest to me too. In several forms. Like using a network mapping tool for e.g. twitter topics such as NodeXL by Marc Smith/The Social Media Research Foundation. Like having ‘live’ network mappings of how distributed conversations I am part of are shaped, such as the images I recently showed of blog conversations, but then interactively. Like visualising the links between posts as Peter went on to do.


Visualisation of blog conversations (a grey box is a cluster of posts referencing eachother


Peter’s visual of links between blogposts


Anjo Anjewierden’s 2007 visual of Lilia’s blog‘s self references on a time axis

For these types of visualisation Anjo Anjewierden as a researcher did some interesting work 2003-2008, such as building those network maps around my blog. He also looked at visualising self-referencing in blogs. There’s just one dimension there, time, he says. I disagree, as linking to oneself is just as much a distributed conversation as linking between others, and Peter’s experimental visualisation above supports that thought. So I’d be interested to see a network map of self references: which blogposts over time turn out to be more central to our writing/thinking/reflection? Much like citings are a metric in academia, they are of interest in the blogosphere as well. Anjo also released several tools as open source if I remember correctly, so some archive digging is needed.

To do what Peter did, retroactively make all the links between my own blogpostings visible, I would first also need to fix the older links. Those older links are strucured differently than more recent ones and now return 404’s. The corresponding posting still exists but has a different URL now.

Posted on
by

Most of my open data work is with government entities to help change their processes, routines and perceptions to ensure steps towards open by design. I almost never really work with open data itself during those activities. So I decided to accept the challenge we ourselves issued with the launch of the Frisian Open Data Platform.

The challenge was to “find out what the planting year was of the monumental tree that is nearest the street light with the provincial ID number 696502”. Finding that out needed to be done by using data from the Frisian Open Data Platform.

Figuring out which data to use was easy. There is a provincial data set that contains the position and ID’s of all street lights for roads where the province is responsible (other roads can be the responsibility of a municipality, or the national government). There is another data set of the city of Leeuwarden that contains all trees of interest within the city limits. If the street light with the right ID is within city limits, it should be possible to answer the question with the tree data set of Leeuwarden.

So what I did was first look in the provincial data set for the right ID. I copied the coordinates that data set gives for that ID into Google Maps, to see where it is on the map, and it turned out to indeed be within Leeuwarden city limits. So the Leeuwarden tree list contains the answer I’m looking for.

Then I started up Q-GIS, which is an open source geo-data viewer (and in fact, a very capable open source GIS *editor* too, as Peter says in the comments). It is possible to connect a CKAN data portal, such as the Frisian platform is, to Q-GIS. Under the menu-option Plugins in Q-GIS one can install a CKAN plugin, which gives you a CKAN logo button in Q-GIS. Pressing that prompts a dialog in which you can specify the right address for the CKAN server you want to use. This was specified on the Frisian platform as https://ckan.dataplatform.nl/api/3/. I also needed to add a default folder that can be used to keep necessary files.

Now I could search within all the Frisian open data platform data sets right within Q-GIS, using that plugin. I first loaded a map of the Netherlands (the TOP10NL map, which is the most detailed map the Dutch Cadastre provides, as a zipfile of 2GB). I used the PDOK Dutch open geoportal for this, for which I had already previously installed the PDOK plugin, in similar ways as the CKAN plugin). Then I added the Provincial street light list, and the Leeuwarden tree list as layers on the map. I then scrolled the map to the location I had previously checked out in Google Maps.

In the screenshot below you see green dots on the red road. Those are provincial street lights. The rightmost green dot is the one we’re looking for. A bit further to the right you see a row of purple dots. Those are the trees, and one of these is nearest our green dot. Now, I visually judged which purple dot is the nearest, although you could calculate it from the coordinates in the data. Also there is some room for error, as most of the trees in that row were planted at the same time as it turns out. By clicking on one of the dots in Q-GIS you can see the data fields and labels attached to it, and that gave me the year of planting.


The map of Leeuwarden, with the street lights as green dots on the red road in the middle, and the monumental trees as purple dots.


At the top you see the depicted map layers (Dutch map top10nl, trees in ‘bomen’, street lights in ‘provfriesland’), below that, when you highlight a specific purple dot, under identification results (‘identificatieresultaten’), PLANTJAAR is the field with the year of planting.

Posted on
by

At Re:Publica in a session on data visualization to make sense of globalization, the release of a very cool dataviz project was announced for next week: The OECD Regional Well-Being Index. ‘Truth and beauty operator’ Moritz Stefaner, who contributed to the visual aspects, made this announcement during the session and gave a sneak preview.

It is a follow-up of the OECD Better Life Index (also very cool), and a new incarnation of the statistical regional explorer.

What it allows you to do is explore regional data, on the basis of what you deem relevant, and then find out which regions in other OECD countries have similar profiles. This is important, as until now OECD data was mostly presented on national level, but the more profound differences are usually found within a country, or when comparing regions, not countries.

If you do such a comparison for Berlin, as shown in the pictures, you find out why Peter Rukavina likes Berlin so much: it is statistically similar to his home Prince Edward Island, just more urban and with a wider variety of things on offer.


Berlin, with Prince Edward Island mentioned as similar region


PEI, statistically similar to Berlin

The existing OECD Regional Well-Being Index is already a great and beautiful project. It moves away from ranking countries, as that has no real meaning (in the sense of scope of interventions or policy consequences). You can create your own set of important indicators, and your choice as well as those of other visitors is used again as data to improve the visualization of the project itself. The top layer of the index is playful, and doesn’t throw all of the statistics in your face at the beginning. If you want you can dig much deeper and get much richer detailed numbers.

For more OECD data visualizatons see their Data Lab. Also check out the dataviz portfolio of Moritz Stefaner, who created the key elements of the OECD visualizations.