[This a guest post by Roxana Torre*, explaining the process behind the creation of the ‘How liveable are cities?‘ visualization.]
The big question of the contest was “Where is the best city in the world to live?”. Participants were asked to create a new liveability index based on at least one of the datasets provided and additional data and to create a visualization. The two datasets provided for the challenge were the Worldwide Cost of Living index and the existing EIU liveability index based upon 30 factors spread across five areas: Infrastructure, stability, healthcare culture and education.
What is the liveability index?
It’s easy to think about the liveability index as a measure of how good or bad living conditions are at a certain place. However when we think about good “living conditions” there are a number of basic needs and circumstances which most of us will consider as important, while other are more personal. In order to have a better idea about in which direction I had to look for other factors it was necessary to know about the background of the index and in which way it’s being used. According to EIU’s description:
“The survey originated as a means of testing whether Human Resource Departments needed to as-sign a hardship allowance as part of expatriate relocation packages. While this function is still a central potential use of the survey, it has also evolved as a broad means of benchmarking cities. This means that liveability is increasingly used by city councils, organizations or corporate entities looking to test their locations against others to see general areas where liveability can differ.”
The question for myself was which other factors can make a difference when you think about living in a city. I was born in Buenos Aires but I’ve been living in Delft (the Netherlands) for more than 20 years. For me, it’s easy to see how different it can be living in a big city versus living in a small one, living in South America vs living in North Europe. The conditions in both places are quite different in a lot of senses, both having advantages and disadvantages.
From the comparison of these two situations, different possible factors came to my mind:
- Availability of free time: this is a very important factor which was not taken into account in the original index. A beautiful city where you have to spend most of your time working or commuting is for most people not so attractive.
- Leisure possibilities: these are included in the actual index as access to cultural events, but I think that the access to natural places is also very important (distance to beach, mountain, natural parks)
- Environmental issues: how is the air quality in the city and the availability of green spaces
Other factors which could be taken into account are weather, hours of natural light, etc.
Going a bit further, if we consider the original purpose of the index of assigning an allowance for expats, an important factor could also be which are the possibilities of integration into the local community.
I also thought it was a good idea to introduce a more subjective indicator. It’s clear that there are societies where people are happier than in others and this doesn’t directly relate to the other mentioned factors.
Data and licenses
With these ideas in mind I began looking for data. The first problem I found was that it was quite difficult to find appropriate data for all the cities or at least a representative part of it. Some relevant research was limited to countries or continents, or just a small set of cities. The second huge problem were the licenses!
According to the “contest official rules” , by entering a submission in the contest, the participant should grant (among others) the right to use it for commercial purposes. This implicitly means that the datasets to be used should be available for commercial use, which is mostly not the case.
Many of the datasets I wanted to use were only available for “Non-commercial use”, this made the task of looking for data quite complicated. For some of the datasets the license was not directly clear. The possibilities were therefore to use open data available for commercial use (for instance with a public domain license) or to ask for permission in the other cases.
In this way I managed to collect a few datasets which I used to create two extra indicators: environment and life satisfaction. The environment indicator is the result of a combination of air quality data with green spaces availability. Unfortunately there was no green spaces data for Australia, so these cities were not included in the new city ranking. I also obtained permission to use the life satisfaction index. More details regarding the data and sources can be found at the submission description.
Of course an alternative could have been to collect the data myself but I didn’t consider this as a feasible task as there were almost 140 cities in the list, just over one month to prepare the submission and only individuals (no teams or organizations) were allowed!. Here I must say that Filippo Lovato (the author of the winning entry) did an admirable job by gathering most of the data himself for 70 cities.
For the visualization there were different factors which I thought should have to be clear at a glance. On one hand we are talking about “rating” cities and therefore the sorted cities going from most livable to less livable had to be part of the visualization. On the other hand the individual magnitude of the liveability index had to be visible as well. Using an icon for each city and relating one of the icon properties to the magnitude of the index seemed to be a normal approach in this case.
But because the liveability index is the result of a weighted combination of 5 different indicators, I wanted to make the magnitude of these indicators visible in the icon as well. In this way the visitor could see which are the strengths and weaknesses of the city. I tried different (more and less abstract) forms but finally decided to use a traditional but simplified radar chart. This kind of chart was a good solution: the different axes show the different indicators and at the same time the surface generated by joining the points in the different axes with lines gives you an idea of how livable the city is.
To show the rating in both situations (existing Liveability index and the new one) the cities have been sorted going from most livable to less livable.
Giving context to the data
An important part in my process of creating a visualization is to provide some context for the data so that the users can take their own conclusions.
In this case I gave a different color to the cities according to the continent which they belong to. This gives information about where the most and less livable cities are situated.
In order to facilitate finding correlation between liveability and other factors, I introduced the possibility to sort the cities according to three different factors:
- Human development index: The Human Development Index (HDI) is a way of measuring development by combining indicators of life expectancy, educational attainment and income. The index is country wide.
- Footprint: Ecological footprint “represents the amount of biologically productive land and sea area necessary to supply the resources a human population consumes, and to assimilate associated waste.” (see more here)
What becomes visible?
Looking at cities’ radars in the resulting visualization, some interesting aspects come to light (and that’s exactly the reason why I’m doing this kind of job!).
You can see the preponderance of yellow and orange at the bottom of the visualization which means less liveability for cities located Asia and Africa, but you can also see that there is some yellow at the top (Osaka, Tokyo, Hong Kong)
If you concentrate in the top axis (stability) you will see that most South American cities score quite bad, while scoring relatively good for the rest of the items.
In the new liveability index it’s easy to see that a lot of cities score quite bad in environment. However, we cannot say that cities located in a certain continent have more environmental problems than in other continents, Bogota and Quito score good in environment while Buenos Aires, Lima and Montevideo don’t!
Another interesting finding is life satisfaction: looking at Hong Kong you can see that even when this city is pointed out to be the best city according to the winning entry, people living there is not very satisfied! This is confirmed by the reaction of people to the winning entry (see comments). It’s also visible that in different Latin American countries, life satisfaction is relatively high. Mexico is a good example of this.
By sorting the cities according to population, it becomes clear that liveability is not really related to population. There are cities with a high liveability index between the most populated cities as well as between the less populated
Obviously there is a correlation between the HDI and liveability as we could have expected. How-ever an interesting fact to look for here, is that there some cities with a relatively high liveability index situated in countries with a not so high HDI, and the other way around.
And what about (ecological) footprint? By looking at the cities sorted by footprint, you can easily see there is an inverse correlation between livability and footprint. The more livable the city is, the bigger the footprint, this is not a surprise because better facilities mean more consumption. Here it’s interesting to see that some cities in South America have a relatively high liveability index while having a low footprint (Lima, Bogota, Quito, Buenos Aires). The same happens with some Chinese cities.
Because I’m mostly working at the same time in other projects when preparing a contest submission, time is limited (and the time in this case was already short!) so it’s necessary to take decisions to be able to deliver a finished product which makes some kind of sense.
*Roxana Torre is a media designer (MA) graduated at the Piet Zwart Institute in Rotterdam and has a background in land-surveying. Since 2000 she runs her own studio in Delft, The Netherlands. She combines analytic design and programming skills to transform large datasets into interactive data visualizations. (www.torre.nl)