Interactive Visualization of Urban Health and Well-Being in Big Cities
This website is part of the course Geo 878: Geovisualization . As our final assignment, we created a web page, which shows the progress we made throughout the course. Thematically, we focus on urban health, more specific, we created a City Health Index. The page starts off with some useful background information about the topic, the creation of the map and the data we used for the project. Then, the project section shows the interactive web map solution, and answers some possible research questions. For interested people, there is also a section with further reading about the topics and the methods used for the project. Finally, the website gives information on the team behind the project, and their contact information.
According to the World Health Organization, the concept of health includes three components: (1) the social and economic environment, (2)
the physical environment and (3) the persons' individual behaviours and characteristics (World Health Organization : 2019). Two out of these three
components are related to the persons surroundings and living conditions. Today, about 55% of the world’s population lives in cities, a number which is
expected to rise to 68% by the end of the century, making urbanization one of the key challenges of the 21st century. (United Nations Department of
Social Affairs, Population Division : 2017). For this reason, the development of healthy and sustainable cities should be a priority for policy-makers.
Cities provide great opportunities for families and individuals to earn a living and attract a number of people with their rich offer of cultural events, transport and job opportunities. Despite the fact that health is one of a city’s biggest assets, most of the urban population suffers from inappropriate housing and transport, bad sanitation and waste management and air quality that does not meet common health standards. Most of the top ten deaths are related to bad urban design (World Health Organization : n.Y). Top death causes are known to be related to bad air quality (which leads to heart problems, strokes, chronic respiratory diseases, lung cancer and pneumonia amongst others), or the shift from rural to urban living (World Health Organization : 2018). This includes the health effects from switching from fresh to processed food, and the decrease of walking in a car dependent city.
For these various reasons, the World Health Organization has launched the Urban Health Initiative, with the goal to help low and middle income cities develop policies which benefit the population’s health. These cities especially face problems, as their population increases the fastest, and outgrows a government’s capacities to plan and develop healthy infrastructure (United Nations Department of Social Affairs, Population Division : 2017). This initiative is closely related to the United Nation’s Sustainable Development Goals, more concrete the following ones (WHO: n. Y.):
Due to its complexity, the topic of urban health includes a number of possible research questions and topics.
When narrowing down the possible solutions, we wanted something that covered many aspects of health. Therefore, we decided to create
our own health index. This would then allow us to compare the different cities’ health.
One next step was the choice of the cities that we would compare. After a some data research, we decided, to focus on the European Union (and Switzerland) and select the 14 largest capital cities of Europe, as well as Zurich. The main reason for this was the availability of data. Health data are often not available on city level, let alone in developing countries. The European Union’s statistical authority EUROSTAT collects the most important socioeconomic factors on city or metropolitan area level. Therefore all of our data originate from the same source, which guarantees the highest level of comparability possible. The downside of this that the approach does come across as a bit Eurocentric, and not available for the biggest cities in the world.
When researching for the project, we found a number of indices measuring city health. The methods of our index are based on a publication by the World Health Organization called: The Urban Health Index: a handbook for its calculation and use (World Health Organization : 2014), which is a well justified framework for creating a health index. In their framework, the authors identify five categories which are determinants of health: Health, Environment, Geography, Economics, and Sociodemographics. We decided to use these same categories and, use at least one indicator for each one of them. The following list gives an overview over the data that was used in the index. Unless stated otherwise, the data originates from EUROSTAT and was measured on a city level.
Most of the values depend on a factor which is related to the size of a city which is not suitable for the calculation of an index. Therefore, most values had to be normalized between 0 and 1, unless they were already a ratio in that magnitude (number of university students, proportion of working population). Furthermore, the impact of an increasing number had to be considered. For some indicators, a high number means a positive effect, while for others, a high number means a negative effect. For those indicators where an increasing number signifies a negative effect we subtracted the number from 1, which inverts the relationship, making higher values more positive.
For the project, we used data from two main data sources. For the areas of Health, Geography, Economics and Sociodemographics, data from the European Union was used. Their statistical authority is called EUROSTAT, and they provide the official statistics of the European Union. Our choice included Zurich (because it is our hometown), which luckily was also included in the statistics. This one source guarantees that the statistics are as comparable as possible. Furthermore, having the different indicators in already in one dataset each, made the task of collecting all the statistical numbers needed much easier. The downloaded data included the years 2009-2018. Out of these the years 2010 – 2014 were selected because they were the five year period with the least missing values. For the category Environment we collected data from the OpenStreetMap API. This allowed us to download the areas free of charge, to calculate the relevant metrics. Again, having the same data source for all the cities is an important part in making the different cities as comparable as possible. For the category of the environment, the temporal aspect was not considered and the same value was used for all the years. However, we argue, that this does not affect the index by a great deal, because cities usually consist of heavily built areas and parks are rarely given up. After the data was downloaded from the respective source, it had to be pre-processed, before it could be used in the shiny app. This especially concerned the EUROSTAT Data. One of the major problems we encountered was the missing data. Many cities seem to not collect all the numbers every year, or decided to not report them to EUROSTAT. However, for the calculation of the City Health Index, we were reliant on having the data for every year from 2010 - 2014. There were three difference procedures on how we handled missing data, depending on how much of the data was available.
An attempt to measure and visualize urban health in cities in Europe
The first step was to set up the basic webpage for the project. There were a few requirement on the structure of the page.
The homepae should at least contain the following elements: a homepage, with a picture and a title, a background section, a link section with further reading,
a contact section about the team members, a method section with the output of the labs.
The webpage uses basic HTML technology, as well as CSS files that we found online with W3Schools. We decided to use already available stylesheets which would allow us to focus on the development of the map application and not on designing the web page. Apart from that, we came up with our own way of structuring the webpage. We decided to create an about section with all the background information on the data, the topic and the concept. Further sections include the project with the actual interactive web map, a lab section with the result of the lab, as well as a further reading and a contact section. We decided to keep the structure as simple as possible, which guarantees a high level of accessibility. Therefore, we chose to keep a single one page layout, which guarantees an easy access for the vision impaired. Furthermore, some design considerations had to be made. We decided to keep a clean functional layout with a neutral grayscale color scheme. Further colors will be added in the web map as well as in the graphs of the data exploration.
After setting up a webpage, we explored the data that we are using for our application. Read more on the data used in the corresponding data section. As a first step, we loaded all the EUROSTAT data into R and did some visualizations to get an overview about the dataset. We realized, that some years were missing, therefore, we decided to visualize the years, in order to gain an overview on the data.
The visualization shows a plot for every city, with the symbols showing the different indicators While the information is not as clearly visible, however, shows the number of records per year, and that not every year has the same number of records. Therefore, we decided to narrow the years displayed in the application down to a 5 year period.
We looked at the numbers in a bit more detail, and also generated a plot with the number of entries per year. This allows us to see which years have the most entries and is a good foundation for the choice of the timespan that is covered in th application.
The maximum number of records possible is 90 (six indicators and 15 cities) which was not achieved by any year. The year 2011 had the most records with a number of 78. Based on the numbers, we decided to use the time span 2010-2014 for the analysis in the application. Partly, this is due to the fact, that there were many records in this time period, and on the other hand, the start point (2010) seems like a logical break because it is the start of a decade.
After looking at the years, we also decided to look at the indicators and how many entries they each have. We also counted the number of records per indicator. In total
we had data of 15 cities over a time period of 10 years (we looked at the full dataset, instead of just the five year period from the paragraph above), therefore the
maximum number of records possible is 150.
The indicator of population had the most entries with a number of 113 entries. Most of the missing values were 2016 to 2018, which doesn’t concern our analysis. One indicator with very few entries is mortality due to pulmonary disease, which is why we decided to look at it in more detail.
Taking just one indicator gives a better view over the distribution of the values. It should be noted, that the values are not standardized, therefore
looking at them does not allow to make any assumptions about the health of the city. However, we can see, that for most cities, there is a linear relationship
between the indicator and the time (this, to some extent can also be observed in the first visualization). Therefore, we decided to fill the missing values
with a linear interpolation. This gives a good approximation of the value.
After this linear imputation, we realized, that Vienna had no value for the two indicators mortality due to pulmonary disease and mortality below 65 years. A quick look into the metadata shows that the city of Vienna treats these data as confidential. Therefore they are omitted in the calculation of the City Health Index.
The next step involved the creation of an interactive web map solution wich displays the actual index. The application is based on R's shiny package. For a view of the app and a discussion of the results, please refer to the project section.
This section gives an overview of the data, literature and ressources used for this project
Two MSc Geography Students from the University of Zurich
Please contact us for further information or questions!