Hackathon Elections a Great Success

13 March 2017

Last week a range of media reported that 75% of eligible voters were still hesitating between two or more parties for Wednesday’s Dutch Parliamentary elections. With this in mind, last Friday some 60 participants came along to Utrecht for the Hackathon Elections, hosted by The Analytics Lab and Cmotions, to take part in the second edition of the event. Eleven teams from the financial, energy, automotive, insurance and other sectors endeavoured – despite the still widespread hesitation amongst the electorate – to predict the distribution of seats from the forthcoming elections as accurately as possible.

The hackathon was kicked off at 14:00 by the organisers who had reinvented themselves for the occasion as the leaders of parties with such original names as the Partij van de Analyses [the Analysis Party] (PvdA), Volume and Value of Data (VVD), Correcte en Degelijke Analyses [Accurate and Reliable Analyses] (CDA) and the Statistisch Georganiseerde Partij [Statistically Organised Party] (SGP). Surprisingly, the Partij Voor de Vrijheidsgraden [Degrees of Freedom Party] (PVV) was absent from the debate. Once the polls had opened, the participants frantically set about analysing and modelling. They used well-known techniques such as Random Forest Modelling and Neural Networks, but creative types also got to flex their mental muscles a lot too. One of the teams based their predictions on data from the Kamergotchi app, and the team from RDC drew on their knowledge of the Dutch car market with the “Krol Correction”: the phenomenon of older people tending to buy increasingly trendy cars and thereby distancing themselves from the “50-plus” label given to them by Henk Krol, the leader of the senior citizens’ party. The correlation between the number of red cars sold and seats for the SP, called the “SP Coefficient”, was invented by Team RDC.

At 17:30 the first results started trickling into the organisation from different parts of the country and the votes were counted. Once the Marketing Association of the University of Groningen, MARUG, handed in their predictions with just a minute to spare before the deadline (you can tell they’re real students), the team captains prepared themselves for the big leaders’ debate. In the debate, the participants were asked about the tactics and modelling techniques they used. Despite the wide variety of sources and analytical methods used, the participants were more or less agreed on the distribution of seats, with a maximum spread of seven seats for the different political parties. The biggest outlier was the number of seats MARUG predicted for the PvdA. Despite the PvdA being opposed to reinstating the basic grant for students, the students allocated the party as many as 29 seats in Parliament.

When the election results are revealed on Wednesday, a “Root Mean Square Error” method will be used to work out which team is the winner of the second edition of The Analytics Lab Hackathon and will lift the prestigious gavel.

Watch the Aftermovie here (with subtitles)!