Presidential elections have just finished in France, with two rounds on April 23rd and May 7th 2017. The results of the elections are available online for each ~67000 polling stations around the country. This data can be used to gain insights about both the electorate and the candidates. Examples of basic questions worth looking at are:

• How are candidates' scores correlated with each other? Can you infer 'political affinity' of the candidates just looking at the data, without any other a priori knowledge?

• How are scores of various candidates correlated with the turnout? Supporters of which candidate are the most 'participative'?

• How did the votes get redistributed from the first to the second round? Can you build an a posteriori predictive model of the second round results taking as an input the results of the first round ?

• The data provides coordinates of each polling station. Can you gain any insight from the geography?

Some preliminary analysis is described in two blog posts here and here. This dataset is inspired by an analogous one for the 2016 US elections by Ben Hamner.