In a previous article, using Big Data and Data Mining, the staff of The Current Politics tried to predict the outcomes of the Dutch general election that took place on March 15, 2017. The voters selected the 150 Dutch representatives in the lower — and law-making — chamber of the Dutch parliament. In another article, our staff used psychology to predict the coming French presidential election in April and May 2017.
The Big Data and Data Mining concepts are related and simply refer to the large-scale computer processing of the extensive and detailed amount of data. In our case, the data was a number of searches used in the Google search engine. Our analysis pivoted on the assumption that voters would most probably search for things they like; especially in the election times when people would search for promising election platforms.
So, who are the biggest winners?
To begin with, the Netherlands has a proportional representation voting system, which means that every vote counts, and also means that any party who pass the threshold of 0.67% of the total vote will be represented in the parliament. This would significantly increase the number of parties within the house walls. It would also increase the probability of the appearance of far-right parties compared to first-past-the-post system. The following chart represents the seat allocations:
On the other hand, the predicted results using the Big Data were:
In the above predictive charts, the interest in the CDA party was showing the maximum potential. The VVD was a little ahead of PVV. Meanwhile, PvdA party was roughly the same level of interest as the GroenLinks and D66, while SP party had the minimum interest and does not appear on the chart.
The predictions showed — with exception to CDA party — it was a very tight race, and that was a good point and agreed with the polls at that time. The reasons for the discrepancies are mainly two, which should be taken into future consideration: any pre-election crisis and the abbreviations.
Any pre-election crisis
March 9-11, a Netherlands — Turkish crisis appeared on the surface of the Dutch election. Turkey wanted to mobilize its citizens within the Netherlands to approve a Turkish constitutional amendment that would grant Mr. Erdogan sweeping presidential powers. The Dutch authorities with Mr. Rutte’s leadership refused any rallies and expelled a Turkish minister and prevented another from landing in the Netherlands. That quick and responsive nationalist action may have pushed many undecided voters — up to 40 % — to make up their minds and vote for the VVD — Mr. Rutte’s party.
VVD, PVV, PvdA, D66 as abbreviations have very concrete political meaning in the Dutch political system, this is not the case for CDA, SP, and GL. In future analyses, proper exclusions should be applied and the most famous abbreviations used within those future cases should be thoroughly looked into and properly grouped and analyzed.
To give another contemporaneous example about the prediction powers of Big Data, the following charts show clearly the edge held by Ms. Hillary Clinton against Mr. Bernie Sanders during the democratic nomination at the primaries’ stages. Another prediction was the obvious edge — that went undetected by most U.S. opinion polls — by Mr. Donald Trump over Ms. Clinton during the U.S. presidential election.
The opportunities and potentials for using the Big Data and the Artificial Intelligence to unlock and foresee the political and the economically intertwined uncertainties are limitless and encouraging.