Long elections, long forecast
In 3 years of analyzing social media for forecasting the people’s will and monitoring the situation development in different regions of the world we usually worked with hours, days, rarely — weeks — until the «hour X».
Now we decided to use our solutions and analytical methods for a long-term important event — elections in France, that will happen in spring, 2017.
Data for May 2015 (May 23-25):
- Mentions of politicians: 41 948
- Unique authors from France: 16 755
Active negative sentiment towards Hollande was detected in social media for example a tweet (and publication) of LeParisien: «Présidentielle : 77% des Français ne souhaitent pas que Hollande se représente
(«Presidential ellections: 77% of French do not support Hollande») was «supported» in social networks by more than 2 thousands people and caused an active discussion— more than 400 comments have negative sentiment towards the politician’s actions.
The most popular (quoted) «political» tweet is a musical remark by LauraMounsi__ — 660 retweets:
(«Rihanna had Obama in her music video, but I cannot imagine that someone from top french celebrities used an image of Hollande for his/her music video»)
Daily forecast is an average between two scenarios described below: inertial and dynamic that set the borders of forecast corridor. Election results forecast at May 24th is the following:
- Bronislaw Komorowski – 51,69%
- Andrzej Duda – 48,31%
96% of Russians support the results of the referendum in Crimea.
Here are the results of social media users’ opinion regarding the outcome of the referendum in Crimea that took place on March 16th, 2014
- Support the results: 96%
- Do not support the results: 4%
Research commenced at 22:00, March 16th and finished at 10:00, March 17th 2014. During this time, 188 874 messages by 104 965 Russian users were analyzed.
France. May 2012.
During the president elections, we analyzed the whole message stream in social media and using our self-developed methodology made an elections outcome forecast.
Analysis base: amount of unique social media authors who shared their opinions about the candidates — 103 thousands
USA. November 2012.
As a first experience, those results exceeded all our expectations and elections in USA became our next target. Analysis base: amount of unique social media authors who shared their opinions about the candidates — 5.7 mln. Given the features of USA electoral system, our starting objective was to forecast the result based on the electors ratio.
(by the electors ratio)
Of course, those inside experiments, other than giving us an ability to accurately define the actual priorities of modern social opinion study development, also gave a solid base for further work directed at creating the new scientific apparatus for analyzing the new reality before us.