Tuesday, January 10, 2012

Power of community analytics and predicting elections

Parliament Building in Reykjavík, Iceland
Image via Wikipedia
I have written about Clara and their Icelandic PR and Brand monitoring product Vaktarinn. What fascinated me while we were discussing the investment in Clara was a couple of projects that team had done with the technology platform that they had created to analyze community dynamics. Actually, team Clara had blogged about it but  it was all in Icelandic and of course I had to learn from the team.

National Assembly 2009
Right after the collapse of the Financial system in Iceland and before the fall of the then Government, there was a National Assembly organized and it was called Þjóðfundar in Icelandic. This was a cross section of over 1000 people from all walks of life in Iceland and there was a crowd sourcing movement to determine the values  that were important to Iceland as a nation. The values list is published here. Clara's community analytics platform was used on the captured words, discussions and debates.

The team then used the data from National Assembly and took all the speeches made in the Parliament of Iceland by the elected members. They mashed up the data to see the correlation between the words used by the Members of the Parliament and those used by the crowd in the National Assembly. The picture to the right shows the names of the top 5 MPs with the highest correlation and bottom 5 MPs with the lowest correlation. The one with the highest correlation became the Finance Minister of Iceland and 5th highest became the Prime Minister. What is even more interesting is that this analysis was done before the General election. The blog link that I had provided before to the analysis has lots of graphs, the background and description of the analysis. One political party scored very high in the correlation and that was the party that showed the highest growth rate in the number of votes in the general election and took many of the important Ministry appointments. Oh! what one can predict if one had the right tools and data, especially unstructured data.
Enhanced by Zemanta

No comments: