Intermediate Implementations: First Post

After a fairly long hiatus, we are back in another brief blog post to update everyone on how we are doing as we implement our project. Indeed, actualizing the ideas we converged on — a process which we described in more detail a few weeks ago — has been a challenge, as it involved actually acquiring new skills and familiarizing ourselves with new tools and languages. Learning to work with these tools, some of which are mentioned below, can be time- and energy-consuming, which explains our absence. In any case, let us now provide a snapshot of some of the objects we have begun to create, to see where we currently stand.

First off, we (all credits go to Luka) thought it might be neat to create a dashboard for us to display our work on the web. To achieve this, we opted for R shiny, an R package designed to accomplish exactly such tasks. Our preliminary dashboard can be found in its entirety here. Here’s an overview of what it looks like, with various figures and variables of interests occupying different frameworks and tabs. Most of the plots are done in plotly with some in d3, but this is just a preliminary step to better grasp the possibilities available to us, as we are in the process of transferring the plots to Vega.

A broad overview of our R Shiny dashboard

Let’s take a closer look at the interactive web app enabled by R Shiny, and what figures we decide to display. In the first tab, we show a choropleth map of Europe, which is the geographical focus of this app for now. The map shows countries’ Freedom House political rights score (ranging from 1 to 7 in discrete values), and suggests a gap between Western and Eastern Europe, the latter rarely achieving the best possible score of 1. The Baltic states stand out in this domain, as they obtain results on par with Western Europe in terms of political rights. Hovering over any given country with the mouse gives the user its exact score on the scale.

One can consider this sociopolitical data in conjunction with the economic data visualized in the second tab of our app, an an early attempt to address our primary research question. This second section shows time series GDP growth data as a line graph, with a brief description of the metric to the left.

For now, we consider two alternatives on how to display the GDP data. The first option, shown at the top reveals country names as the user hovers over each line with the mouse. The second option is to select the European country of interest using the drop-down menu to the left. The top graph is created in d3.js. Interestingly, countries that fared less well in political terms have had higher growth in the past two decades relative to countries like Germany, or France. During the recession of 2009, the aforementioned Baltic countries took the biggest hit, and countries that displayed the most economic resilience were Albania, Kosovo and Poland, respectively. Another interesting observation to glean from the top chart is Greece’s struggle in the wake of the global financial crisis, with low growth for an extended period of time, as well as Ireland’s spectacular growth in 2015.

In the first bar chart of the third tab, we try to combine countries’ political rights rating — captured by color variation — with the percentage of their parliament seats occupied by women. Countries with the highest rates of women in parliament fare well on political rights, being almost exclusively purple (the best score possible on the scale). However, as women’s representation drops, so does performance on the rights index, with bars become increasingly colorful. Typically, Western European countries have higher women’s representation. Serbia stands out as an outlier, though, with a low political rights score yet good gender parity, representationally-speaking.

The line chart in the upper right corner is a time series similar to that of GDP, while the figure on the lower right is an early attempt at illustrating women’s representation in parliament.

As we explored the data, we tried to see if there was any relationship between a country’s level of inequality (measured by its Gini index) and its provision of political rights. There is no clear conclusion to be drawn at first sight; however, the country with the highest inequality (Turkey) also does the poorest on the political rights scale. Similarly, countries with the lowest levels of inequality score well on political rights.

We also created an early prototype of a radar chart in d3, which allows us to break down the Fraser Institute’s Economic Freedom Index into its sub-components, while comparing two countries at a given time point. This makes for an interesting interactive experience. Below, we show an example of Chile versus Albania. We will try, in the coming weeks, to further perfect this visualization which is already in good stature. Because none of us having coded in Javascript before, this was not an easy task.

All in all, it has been a process to learn to work around some of these packages and tools. However, we have some early figures to show for, and as showcased above, can draw some early observations from these visualizations, which broadly highlight interesting trends.

4 students enrolled in the master’s program of statistics and data science eager to start blogging on Medium