Open Data Zurich offers data of the city's tree cadaster. In order to showcase the application of so-called small multiples in visualisation, I generated several iuxtaposed graphics using this data set.

Small multiples (a term coined by Edward Tufte; sometimes also called trellis or lattice charts) are a series of similar graphs using the same axes and scale. This allows the individual graphs to be easily compared to each other. Small multiples are thus an effective way to visualise different partitions of a dataset.

Small multiples are often used in information visualisations where the building blocks are charts. As this example shows, though, small multiples can also be built from maps. In this case, the small multiple aims to visualise the species-specific spatial distribution of individual trees within the city's territory.

→ Blog post “Informationsvisualisierung: Small Multiples”

→ Small multiple, low-res, analytical version (png)

→ Small multiple, high-res, analytical version (png)

→ Small multiple, low-res, designed version (png)

→ Small multiple, high-res, designed version (png)

→ Small multiple, high-res, designed version (pdf)

Strava is a popular fitness-tracking platform for avid cyclists. In this analyis, I compare crowdsourced Strava track density with the existing bicycle infrastructure (lanes and paths) of the city of Zurich.

The infrastructure data is taken from Open Data Zurich. In the visualisation, the infrastructure is categorised into unidirectional infrastructure (lane or path only on one side of the road/street) and both-sided infrastructure.

I produced this visualisation for a discussion of the merits and potential pitfalls of crowdsourced data, or VGI (volunteered geographic information). If this topic interests you, you can find the accompanying blog posts here:

→ Zurich Bicycle Infrastructure vs. Strava

→ Blog post “How can I use crowdsource data?”

→ Blog post “Wie kann ich Crowdsourcing-Daten nutzen?”

GeoHipster is a US-based community website of self-professed GeoHipsters (or, at least, geo-enthusiast and geo-professionals). GeoHipster mainly revolves around a series of fascinating interviews with geo-people from all over the world and Twitter conversations. In this project, I mapped the profile locations of the followers of the GeoHipster Twitter account.

Included in the map are people with public Twitter accounts that feature geocode-able information in their profile (in the “location” attribute). A geocode-able location attribute in this case either contains a recognisable place name or lat-lon coordinates in WGS1984, separated by either a comma, a semi-colon, a slash or a vertical bar (this one: |) (e.g: “47.51 / 8.54”).

If a place name is given, it’s geocoded using the Geonames API and a minor random component of usually a few hundred meters is added to reduce clustering of map markers.

A special case are followers who indicate their location as, for example “everywhere”, “global”, “earth”, “somewhere”, “worldwide”, “unclear” or “Null Island”: I took the freedom to assign their location to the latter. It just seemed fitting for GeoHipsters.

→ The GeoHipster Map

→ Some more information about the making of the map.

→ Yet some additional pointers from a similar project.

Both my research and my work have showed me times and times again that working with data requires ethos. Among many other things we need to take into account biases, assumptions, or downright errors in our data and data collection procedures.

In my opinion, as data workers we should absolutely know as much as possible about such limitations of our input and tools in order to take the right steps to either mitigate their influence or – in the worst case – step back from an analysis and look for better data or improve protocols.

Ideally, we can also educate wider society about limitations of data gathering, data itself, and data analysis. In these days, we encounter more and more data, for example in the results of data-driven journalism. Hence, making sense of data and critically appreciating analyses that are carried out will become a skill of growing importance.

I presented The Data Worker's Manifesto at GeoBeer #8 in Zurich. Interesting discussions have ensued. I would welcome also your feedback, if you're interested in a conversation on these topics.

→ The Data Worker's Manifesto (slide deck)

→ Slides with annotation

Switzerland manages to maintain an active beer culture complete with numerous micro-breweries. Inspired by organizing the GeoBeer events, I decided to map Swiss breweries to celebrate this culture.

My initial data source was a Wikipedia list of breweries. I augmented this data with records from the Swiss federal tax authorities (an alcohol tax is levied on commercial brewers). Additionally, I decided to base the back-end of this web map on a Google Spreadsheet. This means I can let users easily download the data as well as submit their suggestions for additions to the map via a Google Form – a nice two-way path for data-sharing.

Once verified and geocoded, the data is pulled in using Tabletop and visualized using Leafleat with various plug-ins. The code is on GitHub.

→ The Swiss Beer Map

→ The project on GitHub

Part of my work at the Oxford Internet Institute is focused on the geographic coverage of articles in Wikipedias of various languages. For this project we looked at geocoded articles in 44 different language versions of Wikipedia.

Mapping this data shows a very uneven distribution of Wikipedia content over the globe. And this uneven distribution does not portray settled area or population very well either: We find that Western-Central Europe features 50% of all Wikipedia articles analyzed, despite accounting only for 2.5% of the world's land area. The whole African continent contains about 2.6% of all articles while representing 20% of the world's land area and 14% of the world's population.

Digging deeper, we find that as a tendency articles in Western/Central Europe were initiated earlier than articles in other world regions. Also, all European language versions of Wikipedia except English, Russion and Spanish feature more articles in Europe than in the rest of the world – attesting to the self-focus bias of Wikipedia.

You can find a much more detailled account of the methodology and the results (including more graphs) on the project page.

→ The Geographically Uneven Coverage of Wikipedia

This project visualises the terms Google Autocomplete associates with different countries. The results show how Google can actively shape the knowledge we obtain about different parts of the world.

→ The World Through the Eyes of a Search Algorithm

→ What Google Autocomplete tells us about countries

Wikipedia doesn't yet fulfill its promise as a repository of all human knowledge. This map visualises Wikipedia in Africa and tries to understand the observed spatial imbalance of information and knowledge (re)production.

→ Africa on Wikipedia

Somewhere between dataviz and art, The Sky over Zurich displays the color of the sky over Zurich throughout the years. The colours are taken from a web-cam feed (making-of). The Sky over Zurich has also been featured in the Swiss daily NZZ.

→ The Sky over Zurich

→ NZZ article Der Himmel über Zürich

A hexagonal cartogram, a map and a bar plot about population sizes of Swiss cities and cantons, implemented in D3 (making-of). This piece has been featured in a award-winning article series in the Swiss daily NZZ, among others.

→ Swiss Population Cartogram

→ Making-of

→ NZZ article Grosse Städte, kleine Kantone

For the 2013 Cartography and Narratives workshop I carried out an analysis of crowdsourced Flickr content. I posed the following questions:

• Which parts of a city are visited by how many people and covered by how many photographs?
• Do different groups of people document different areas of the city in photographs and thus likely tell different narratives?
• How do photographers and their trails of photographs connect different places in the city?
• How does all this activity vary temporally?

In addition to the scientific investigation and the workshop paper, I produced a visualization that conflates all the data into an animation of a hypothetical 'Flickr Day'. The visualization has been built using Processing.

→ Video 'Diurnal and spatial patterns of georeferenced Flickr photographs in the city of Zurich'

→ more background information

A simple thought experiment. Imagine: Switzerland as one of the world's city states. What area would the population occupy at the density of, say, Hongkong, Singapore, Manhattan?

→ Switzerland as a city state

In this project I used semi-automatic Twitter community discovery approach to find and map the network of Switzerland-based Twitter users interested in GIS, web mapping, and cartography along with some additional analysis. The results have been published on my company's blog (in German) and in a journal (in German, French, and Italian).

→ GIS-Netzwerk im Zeitalter von Social Media (in German)

→ FGS/PGS article La scène suisse des SIG dans l'espace virtuel (in French)

→ FGS/PGS article Lo scenario svizzero dei SIG nello spazio virtuale (in Italian)

Using a list of 300+ Twitter accounts of German-speaking journalists, compiled and crowdsourced by David Bauer, I used the Twitter API to build their network. Mapping it in Gephi, revealed some structures in this extraordinarily dense network.

→ Journalists' Twitter network

For this visualization I tinkered a bit with the Processing framework. This resulted in a small two-part introduction to visualising spatial data in Processing. The end results are interesting static visualisations of postal code locations (obtained from Geonames) that have been termed 'ZIP scribble maps' by Robert Kosara.

→ Introduction to ZIP scribble maps in Processing

→ the whole series of ZIP scribble maps of Germany, France, Italy and Switzerland

After the devastating nuclear catastrophe in Fukushima, Japan, I got interested in nuclear power in Switzerland and the associated risks we incur. Hence, in the weeks after the Fukushima meltdown, I carried out this analysis of the population density around nuclear power plants in Switzerland and neighbouring countries. I published a report on the results (in German) that has been cited in the Swiss parliament.

→ Gefährdung der Bevölkerung der Schweiz durch Kernkraftwerke

For the University of Zurich I made a set of 3D geovisualisations of a wind farm and of a terrain parameter. You can see these by visiting the Advanced GIScience course website and refreshing the page repeatedly.

→ Advanced GIScience course website