…and the winner is…. SWEDEN!!! (actually)
Tweet
On Saturday morning I posted this analysis which tried to predict the winner of Eurovision from the Twitter activity during semi-finals. Its prediction was that Sweden was going to win. That part was right. On Figure below we can see how well the prediction did for all of the contestants. Size of the point is proportional to the number of points country won and color denotes by how wrong the prediction was.
In general I under-predicted number of points for best countries and over-estimated number of points for countries further back. Point for Cyprus is not shown as it quite far off (at 4.8). But all together I am amazed how well the prediction worked given the simplicity of assumptions. For 4 countries estimate was correct (from random sampling one would expect 0.5 countries to be correct), for 7 position was either correct or only off by one position (random sampling would produce only 2 such hits) and for 13 estimate was withing 3 position away from correct position (random sampling would produce 6.5). Below are also equivalent Figures for both semi-finals. For semi-final 2 estimate is almost amazingly correct!
Predicting Eurovision 2016 from Twitter data…
[…] last year’s winner was correctly predicted and overall the prediction was quite good (see here for comparing prediction to actual results). The amount of points is much larger than in previous year because of the change to the voting […]