A Geographer
“Cyanomap” was a small application I wrote to automatically compile and map outbreaks of cyanobacterial blooms and other associated news with the help of the Yahoo Placemaker and Yahoo News API services.
Both Yahoo services have since closed,
My probability model of Eurovision Song Contest voting is predicting victory for Azerbaijan in tonight’s contest in Oslo. The model has a high success rate but this year the contest is more open than normal with several countries in with a shout.
Predictions for the top-10 are:
1. Azerbaijan 1.14
2. Armenia 1.56
3. Israel 3.46
4. Germany 3.66
5. Turkey 4.51
6. Greece 5.42
7. Denmark 6.87
8. Ireland 7.97
9. Belgium 14.58
10. Romania 15.48
and for last place:
25. United Kingdom 133.65
…a few hours later…and the winner is…GERMANY…
[youtube https://www.youtube.com/watch?v=UmOeISUYXuI&hl=en_GB&fs=1&rel=0]
Deserving winner but a rubbish performance for the computer model (r2 0.51 for correlation between prediction and result) but at least it got the UK result correct.
The outcome of this year’s Eurovision Song Contest taking place in Moscow tonight, May 16th, will be (according to our computer model):
1. Norway 0.7
2. Greece 1.7
3. Turkey 4.5
4. Ukraine 5.8
5. Azerbaijan 6.4
6. Bosnia Herzegovina 7.0
7. United Kingdom 9.6
8. Iceland 16.2
9. Finland 17.3
10. Armenia 19.6
With the number indicating the country’s “odds” of actually winning the contest.
Norway is a widely tipped favourite, but they have sometimes done less well on the evening than anticipated in the run-up. Out of the projected top-10 Norway has benefited least from “bloc” or friendly voting patterns in the past five contests. Iceland, Bosnia, and Turkey top the list in this regard. Also, in terms of doing well in recent contests, Norway is well below average with only Iceland and the UK with worse records.
Combining track record and friendly voting, indicates a top-4 of Greece, Ukraine, Armenia, and Turkey.
The British bookmakers have a projected top-3 of Norway, Greece and the UK. The UK tends to achieve a top-3 finish when they enter a deserving song. Anything less than a top-10 finish would be a disappointment (and evidence that Europe, rightly or wrongly, really do hate us right now).
Interestingly our word analysis (see below) of recent posts to the 20 top Eurovision sites has the UK and our entry, Jade, as one of the most talked about entrants. This is a feature of “Eurovision week” and had not been the case in previous weeks and combined with the relative surge in bookmaker support, maybe there will be a big surprise tonight… However also keep an eye out for Azerbaijan, Turkey and Ukraine challenging Norway and Greece to the top-2 spots.
The first run of our “Eurovisionomics” model for 2009 indicates that Norway are clear favourites to win the Eurovision Song Contest next month in Moscow. The model predicts another disappointment for the UK with a finish in the bottom half of the results table suggested. The model takes into account various factors including performance in the last five contests and the extent to which countries benefit from “bloc” or friendly voting patterns.
The first run of our “Eurovisionomics” model for 2009 indicates that Norway are clear favourites to win the Eurovision Song Contest next month in Moscow. The model predicts another disappointment for the UK with a finish in the bottom half of the results table suggested. The model takes into account various factors including performance in the last five contests and the extent to which countries benefit from “bloc” or friendly voting patterns.
A tag cloud or word cloud is a visual representation of the relative word content of a website or feed. The word cloud displayed in this post is computer-generated every 6 hours to display the most frequently appearing words in headlines concerned with the ‘environment’ from 12 international news organisations. Each organisation is likely to have some bias in its reporting of environment-related news but word cloud analysis helps reveal the predominantly important issues or stories of the day. Another simple word cloud is implemented at Eurovision Song Contest Voting, where over 100 sites are polled every 30 minutes. These are simple implementations of a much larger project we are currently involved with.
Following Russia’s victory in the Eurovision Song Contest our prediction proved surprisingly correct and was picked up by various media outlets as further evidence of neighbourly voting ruining the contest and the chances of nations like the UK from ever winning again.
The Eurovisionomics analysis had looked at those countries who’d received unusually high scores from the same country on at least four occasions in the past 6 years. Such occurrences were less likely to be attributed to chance and indicate that other factors may motivate voting. Based on this it was concluded that if Russia’s entry was half-decent then their strong showing in this Eurovisionomics analysis should see them through to victory. Further details here.
However the Eurovision Song Contest Voting website has always taken the view that neighbourly voting does not determine the song contest winner. We therefore excluded all votes cast in the 2008 contest that were above the country’s average (+ the standard deviation). This removed former Soviet states votes for Russia and other nations who’d apparently scored a song well above its true value as determined by the majority of the other competitors. The outcome made little difference to the ranked results with Russia still being the clear winner. The exercise was repeated for every contest since 1957 and revealed that without voting bias Russia may well have won in 2007 (they actually finished second).
As for the UK, removing “friendly” votes from the 2008 contest left us with “nil points” – probably well deserved considering we entered a talent show runner-up with a song that managed number 67 on the singles chart.