Euro 2016 Squads Part Deux

Between some of the games I had a go at replicating a plot from liberation.fr on the connections between Euro 2016 players and the country of birth using the circlize package in R. As with the previous post, the colours are based on the home shirt of each team and data scraped from Wikipedia. The values in the parenthesis represent the total number of players born in the respective country, which dictates their ordering around the circle. It is interesting to see just how many players represent countries that they were not born in. Only Romania has a 23 man squad completely full of players born in the country and no other Romanian born players representing other nations.

plot_circle2

I also modified some the my ggplot2 mapping code in my previous post to plot the links between players place of birth and the capital city of their national team: plot_pob1Click on the images if you want to explore. Here is an expanded view to illustrate all the longer distance relationships: plot_pob2All the code is up on my Github.

Euro 2016 Squads

This weekend I was having fun in France watching some Euro 2016 matches, visiting friends and avoiding Russian hooligans. Before my flight over I scraped some tables on the tournaments Wikipedia page with my newly acquired rvest skills, with the idea to build up a bilateral database of Euro 2016 squads and their players clubs.

On the flight I managed to come up with some maps showing these connections. First off I used ggplot2 to plot lines connecting the location of every players club teams to their national squads base in France. The path of the lines were calculated using the gcIntermediate function in the geosphere package. The lines colour is based on the national teams jersey, which I obtained via R using the amazing extract_colours function in the rPlotter package.plot_baseI was not entirely convinced that this plot is too effective as the base camp of each team in France is not particularly common knowledge. This led me to create a new plot with a link from the players club teams to their nations capital city.
plot_capThis shows some clear relationships, such as between the Iceland players and clubs in Scandinavia and clubs in northern England and Scotland with the Irish teams. Here is another plot to show some of the more distant relationships for players from non-European clubsplot_cap2

Finally, I had a go at trying to match my old World Cup circular plot using the chordDiagram function in the circlize package.

plot_circle

I ordered the countries according to their UEFA coefficient, which is based on the performance of club teams in European competitions. Most players playing abroad are based in teams in the top leagues of England, Germany, Italy or Spain. Players based in the English leagues make up most of the squads for teams from the British Isles. There are sizeable numbers of the Austrian and Swiss squads playing for clubs in the German leagues and Croatian players in Italy.

All the code for the scraping the data and producing the plots are on my Github. Check out my next post for more on the players at Euro 2016.

2014 World Cup Squads

I have been having a go in R at visualising player movements for the World Cup. I wanted to use similar plots to those used to visualise international migration flows in the recent Science paper that I co-authored. In the end I came up with two plots. The first, and more complex one, is based on a non-square matrix of leagues system of players clubs by their national team.

gjabelwc2014t3
You can zoom in and out if you click on the image.

Colours are based on the shirt of each team in the 2014 World Cup. Lines represent the connections between the country in which players play their club football (at the lines base) and their national teams (at the arrow head). Line thickness represent number of players. It’s a little cluttered, but shows nicely how many players in the English, Italian, Spanish and French leagues are involved in the world cup. It also highlights well some countries where almost all the players are at clubs abroad, for example most of the players in the African squads.

Whilst the first plot gave a lot of detail, I wanted to visualise the broader interactions, so I aggregated over leagues systems and national squads by regional confederations. This gives a square matrix:

> m
          squad
league     AFC CONCACAF CONMEBOL CAF UEFA
  AFC       49        2        1   3    1
  CONCACAF   0       13        0   0    0
  CONMEBOL   2        0       54  11    0
  CAF        0        0        0  36    0
  UEFA      41       99       37  86  296

The plot of which looks like:
gjabelwc2014r

This type of aggregation works really well to show how few European national players play elsewhere (only Zvjezdan Misimovic in all the European World Cup squads). It also provides a way to compare the share of non-European players plying their trade in the European leagues to those in more local leagues within their confederation.

I scraped the data from the provisional squads on Wikipedia, and then created the images with the circlize package. All the code to reproduce the plots + scraping the Wikipedia squad pages are on the my github.