Colorful Development: Cartagena DataFest 2015

This post has been prepared as part of a submission to the Cartagena DataFest 2015.
Click here to look at the data visualization only on
PREVIOUS: Colorful Development: RGB-coded Multidimensional HDI 
PREVIOUS: Colorful Development: Dynamic Graphs


This post is an update on the Colorful Development: Dynamic Graphs post – a data visualization that has evolved into a full-fledged web-app: an updated, prettier version with a lot of user interface enhancements that now includes the inequality and gender-adjusted human development indices as well.

Colorful Development: Cartagena DataFest 2015

Colorful Development: Cartagena DataFest 2015 – click for interactive

In this post we will look at the evolution of the inequality between the three components (Health, Education, Income) of the Human Development Index (HDI), and the Inequality Adjusted Human Development Index (IHDI) using a web-app I have constructed. To make for a comfortable reading, but also for a strong basis, first we will concentrate on the user interface. I have prepared this as part of a submission to the Cartagena DataFest 2015 data visualization competition. As before, for this, we will need to use tripolar RGBHDI plots (I also call it a colorwheel), a rather peculiar 3 dimensional coordinate system defined in this previous post , so make sure you take a look first, as well as the update discussing converting the RGB plot into a dynamic graph with an adjacent world-map. Read More

The global center of mass of higher education: university rankings mapped

Click here if you prefer to read this post on (2 minute read).
Click here to look at the data visualization only on

It is often-touted that the world has been shifting towards Asia (On all fronts, even Formula 1 🙂 ). Indeed, innovation has clearly gotten a good foothold in the East and higher education has been no exception: in the last 10 years the global center of mass of the top ranked 500 universities has been constantly shifting towards East: 650 kilometers, to be precise. To visualize this, I have created a dynamic map that tracks the top 500 universities and the global center of mass (geocenter) of higher education over the past 10 years!

The global center of mass of higher education: university rankings mapped

Dynamic map of the global center of mass of higher education between 2003-2014 – click for interactive

The geocenter of the top 500 universities in 2003 laid just off of the coast of Portugal. In the past decade, this point (red triangles in the map) has been constantly moving towards East, hovering over the border between Morocco and Algeria by 2014. This phenomenon can be attributed to the appearance of many Chinese universities in the list, as well as some from other Asian countries (Malaysia and Saudi Arabia in particular), and the strengthening of Korean and Japanese entries.

The above calculations are the results of an unweighted arithmetic average of the geographic coordinates of the universities that made the top 500 list, meaning that each university who made list had an equal weight. However, it is fair to calculate the geocenter taking into account the rankings of the universities. When accounting for this (using 1/sqrt(rank) as weights, green triangles in the map) we observe an (expected) shift towards North America, and the United States in particular, where most of the top 100 universities are located, including 8 of the top 10. The weighted geocenter has been constantly hovering further out in the Atlantic, above the Azores, in the past 10 years, but it had a smoother migration towards East than the unweighted one. This means that while it is true that many new Asian entries appeared on the list, they have managed to move up the rankings a bit as well.

The data source for this visualization was the Academic Ranking of World Universities. The data has been processed into JSON format with this IPython notebook, with help from pandas. The universities has been placed on the world map using topojson, after being geocoded by geopy using the MapQuest and Google Maps V3 APIs. the  The visualizations have been done entirely in d3.js and the svg language. The main outcomes are dynamic map of the Global center of mass of higher education between 2003-2014 (static, interactive). If you liked this post or have any questions or thoughts, Like, Share, Comment, and Subscribe!

Religious diversity in Romania visualized on colorwheels

Click here if you prefer to read this post on (9 minute read).
Click here to look at the data visualization only on
A localized Hungarian version of this post also exists.

In this post we will visualize and examine the religious breakdown of the country of Romania and its historical regions. We find that the 4 regions exhibit 4 different patterns and various levels of diversity. A good way to do this comparison is via RGB colorwheels. You can read more about these in previous posts here and here. We will use the colorwheel for world religions presented in a previous post, adapted for the regions and dominant religions of Romania.

As part of this adaptation, we extend the 3-axis, redgreenblue colorwheel to cater for the dominant religions of the country. The result is a 6-axis colorwheel with the sequence of redyellowgreencyanblueviolet. Then, to each of these colors we attach a religion. Using the statistics of the Romanian National Census Bureau, we can define the dominant religions and aggregate the ones with a smaller number of followers. This process yields the following color-coding:

  • Orthodox
  • Catholic
  • Reformed
  • Unitarian | Other protestant
  • Other religion | Atheist
  • Adventist | Pentecostal 

Technical details:

Using this coding, we plot each of the settlements (on a commune administrative resolution) onto the colorwheel. The color (hue) of a data-point (equivalent to the angle on the colorwheel) indicates the dominant religion of the settlement. The brightness of the color (radius) gives the relative dominance when compared to the other religions (There is a caveat when using colorwheels with more than 3 dimensions: while for 3 dimensional colorwheels a 40%-40% share of the two dominant colors gives us an exact indication of the third color, in higher dimensions this needs extra information). All points will lay within the indicated hexagon. For visual guidance, we have shown the diagonals. For points in the vertices, one religion has full relative dominance – 100% of the population follows it. A fully red point indicates a 100% orthodox settlement while a fully yellow point marks a pure catholic commune. Halfway on the diagonal connecting the red and yellow vertices (orange region) lay settlements with a population of half orthodox and half catholic believers. The presence of other religions will result in moving away from this diagonal. When reading the colorwheel, it is important to check the tooltips showing detailed breakdown of the religions and monitor the points placed along the edges and diagonals.

Let us look at the colorwheel of the religions in Romania (interactive infographic, fullscreen):

Religions of Romania Colorwheel

Religions of Romania Colorwheel – click for interactive

The religions of Romania seem to be clustered into 3 groups. It is clear that the orthodox church is by far the largest, followed by a roughly equal number of catholic and reformed believers. There are also smaller groups of other protestant (mainly unitarian) and adventist/pentecostal (mainly pentecostal) followers, while there are almost no settlements with a dominant religion other than these 5.

Read More