BBC Sports Personality of the Year prediction 17th December

As with the X Factor and Strictly Come Dancing analysis, we are analysing twitter data to see if we can predict the winner of the BBC's Sports Personality of the Year.

As noted previously, the X Factor show had a huge number of tweets made about it, with almost a quarter of a million being made on a single Saturday. Strictly has a much lower twitter following, with 25,000 tweets being made in the past week in the run up to the final. The Sports Personality of the Year has even fewer, with approximately 5,000 tweets being made in the last week.

However, we have still been able to undertake our analysis, and the following list has been obtained, in order of likelihood to win:

1. Tony McCoy
2. Mark Cavendish
3. Graeme McDowell
4. Amy Williams
5. Jessica Ennis
6. Tom Daley
7. Lee Westwood
8. Phil Taylor
9. Graeme Swann
10. David Haye

There are some interesting things to note from our analysis.  Jessica Ennis and Amy Williams are very close to each other, and indeed the ordering was the other way around until today's twitter data was included.  It will be interesting to see if the ordering for the girls remains the same over the weekend.  This seems to indicate that there has been a brief surge for Amy, which is likely to be due to the fact that BBC Sport has profiled Amy today.  (Jessica Ennis was profiled over a week ago).

In addition, analysis of the data shows that there was a surge in twitter traffic for Mark Cavendish earlier in the week.  Perhaps this was due to his profile being the first one to be put up on the BBC website (the BBC did them in alphabetical order), but the data shows that his tweets are decreasing day by day.  It looks like the ordering based on tomorrow's data will push Mark further down the list.

We will redo the analysis tomorrow and see if the ordering has changed.

Strictly Come Dancing prediction 17th December

To continue with our theme of analysing reality TV shows, we thought we'd look at the twitter data for Strictly Come Dancing, which has its final show this weekend.

Whereas the X Factor resulted in almost one quarter of a million tweets on a single Saturday night for analysis, Strictly has a much lower tweet count, with almost 25,000 tweets being made over the past week, including last weekend.

Will this affect our analysis? Only time will tell. However clear trends in the data can be seen, and if we repeat the analysis that we undertook for the X Factor, then likely ordering based on twitter data is:

1. Matt Baker
2. Kara Tointon
3. Pamela Stephenson

So this has Matt Baker as the clear favourite.  However if we look at their dancing partners, who are not celebrities, then we see a different ordering:

1. Artem
2. James
3. Aliona

The number of tweets of the celebrities is much greater than that of their dancing partners, so this difference is unlikely to affect the result.  However it does show that if people could vote for the dancing partner only, then it might result in a different winner.

We will repeat the analysis tomorrow to see if there are any late surges by any of the contestants.

Brand Aura founder wins global Technical Innovation award for increased health and safety for mine workers http://tinyurl.com/2uogntv

X Factor prediction 12th December

So we have finally come to the end of the marathon that is X Factor! Cher left the show last night, despite a concerted twitter effort by her fans to encourage everyone to vote for her.

An interesting feature of the twitter analysis that we have done is that certain contestants that have been in trouble have had twitter campaigns waged to try to get other people to vote for them! This has meant that analysis like ours which looks at how close in context the contestants are to winning words will be bias towards those people, such as Katie, Wagner and now Cher. A subsequent reaction is then created however by those not wanting these contestants to win, and we find that the same contestants are closest in negative context to the same winning words.

This perhaps gives us a way to try to analyse the data that we have. Which contestants do the public feel most positive about, overall, given the positive and negative context analysis? In other words, if we add the positive context scores for the contestants in context with winning words, and subtract the negative context scores, what ordering do we get? To put it another way, who do the public love the most and hate the least?!

We have analysed the results from last night's show and also the two previous Saturday shows (i.e. we've ignored the tweets made during the week, and concentrated on the highest tweeting activity which is during the show itself on the Saturday). The results are quite interesting, and can be seen in the plot below.



The analysis shows that One Direction started as favourites a couple of weeks ago. In fact according to our analysis, One Direction have been the overwhelming favourites for many weeks now. However it has been noticeable that their support on twitter has begun to drop in comparison to the other contestants.

This then changed to Rebecca as a favourite, until after yesterday's performances we see Matt now as favourite. This perhaps indicates that this year it is quite an open contest! All of the contestants are different from each other and therefore the individual performances count for alot as to how the public may end up voting.

Our final prediction of the year then, based on Twitter data, is that Matt will win, with Rebecca second. One Direction will come in third.

As always it is worth remembering that this analysis is based on twitter data only, and this may not reflect the actual public vote which has a different demographic.

X factor winner - 3 days to go!

With only 3 days to go before the winner of X Factor is decided, we thought it would be useful to do some analysis on what people have been saying on twitter. Again, we should note that the twitter demographic as we know does not match the demographic of the voting public. However we can look to see if there are any trends in the data, or anything else of interest.

One of the things we can do is look at who is context with the word 'win'. This should give us an indication of who is being talked about, and who the twitter public either most want to win, or who they think is most likely to win.

The chart below shows how each contestant has trended in the past 4 days in positive context with the word 'win'. Here we can see a continuation of the high score for Cher from the weekend. However this did not prevent Cher from falling into the bottom two, and suggests that there is an online campaign to try to return Cher as the winner of the X Factor. We saw a similar thing with Wagner, where he scored extremely well (in terms of the number of tweets in context with 'win' and other winning words).



We can also see a similar story for negative context. In this case, we can interpret the graph as meaning which contestant people most do not want to win. Here Cher is again the top scorer, indicating that people either love or hate her. We see Matt and Rebecca tending to do quite well, indicating that the majority of people like them, particularly in comparison to One Direction and Cher.



These graphs only tell part of the story however. Let's look at some negative context analysis for Cher, can we understand why people do not like her? Or conversely why they feel so passionately about her?

In many respects each of the contestants in reaching this stage is a winner in their own right. Each of them have different singing styles and so the winner is more likely to be determined by the demographic of the voting public. Our analysis will now look to focus on brand loyalty in terms of these contestants - what makes them interesting, and what makes them different.

Let's first look at the words in negative context with Cher. As before, the ~ character indicates a 'not' in front of the word, since it was found in negative context with Cher.

05/12/2010 ~cher ~singing ~ballads ~songs ~done ~slow ~would ~isn ~week ~cocky
06/12/2010 ~cher ~lloyd ~mary ~win ~final ~know ~last ~get ~will ~say
07/12/2010 ~cher ~dont ~win ~want ~2010 ~think ~going ~are ~cant ~watch
08/12/2010 ~cher ~lloyd ~ever ~youtube ~heard ~nickiminaj ~amazing ~week ~has ~fierce

We can see the development of the words over each day. At first, we see the focus on her not singing ballads. We then see comments on the fact that people do not want Cher to win.

If we look at the positive context, at first this makes not so much sense.


05/12/2010 cher mary matt rebecca direction voting leak russia leaked x-factor
06/12/2010 cher lloyd malvern gone tickets gig factor 2010 fixing accused
07/12/2010 cher 2010 going can back cherlloyd follow chertowin lloyd love
08/12/2010 cher lloyd cole cheryl malvern cherlloyd performs vote4cher video factor

There is a large focus here on Malvern, which a quick Google search shows is where Cher is from, and is in reference to her returning and performing a small concert. This story is dominating this analysis, so we can look at the next 10 words in context with Cher to see if there is anything else of interest.

05/12/2010 results will tonight was had are final bottom get cheryl
06/12/2010 show bosses metro booted noonelikesyoubecause direction cheryl ferguson nicki really
07/12/2010 still support were retweet gona damn straight amazing has hashtagged
08/12/2010 awkward let dancing cringe fans mobbed followers twitter gains perform

This time we can see some negative comments come through. Evidently the 5th December shows the discussion of her in the bottom two. Subsequent days however has 'noonelikesyoubecause' coming through along with words like: awkward, cringe. So we see a mixed picture for Cher, which reflects the earlier analysis showing that people either love her or hate her.

We will continue this analysis in the coming days, and repeat for the other contestants.

Public reaction to elimination of Mary from X Factor

Last night's show resulted in Cher and Mary in the bottom two. The analysis of the twitter data had Mary in the bottom but Cher had finished top of our analysis - indicating perhaps that her supporters believed that she was in trouble and so mounted a twitter campaign to try to encourage people to vote for her.

We have seen this before - Wagner often finished top of the twitter analysis (particularly in the week that he was knocked out). This demonstrates how campaigns to try to alter public opinion by the supporters of the respective acts have been fought through twitter.

If we look to analyse the data from yesterday's tweets, we can start with Cher, with the negative context analysis can be seen below. This shows the analysis of sentences that contain words such as: never, don't, not, and so on. To understand the analysis, the ~ character indicates that the word 'not' should be placed in front of the word to approximate the meaning of the sentence.

So looking at the figure we can see that the word ~ballad is close in context to Cher, meaning that there has been a great deal of discussion about the fact that Cher did not sing a ballad for her second song. This was picked up in the judges comments but also struck a chord with the twitter public. We can also see some other words such as ~singing, indicating that there were a number of comments about her "not singing", which may be a reference for Cher to her rapping during songs.

The positive context analysis shown below the negative one, does not have a strong positive or negative sentiment, and rather indicates the discussions taking place that focus on the facts of her being in the bottom two, and perhaps comparing Cher to the other artists.

An interesting thing to note in the positive context analysis are the numbers that come through.  You can see them on the left hand side of the word cloud - 0901 and 104.  These numbers are the start and end of the telephone number that you have to use to vote for Cher.  This indicates a last minute campaign by Cher followers to try to influence the final vote and ensure that Cher does not finish in the bottom two.

If we turn now to Mary, we can again first look at the negative context. Note this does not mean the same as negative sentiment, and in Mary's case we see first of all many tweets referring to the fact that she will not be working Tesco again. We also see some other themes coming through such as ~deadlock, indicating that the elimination process did not go to deadlock, and also some negative sentiment words such as ~better, indicating that Mary was not better than Cher or the other contestants. We can also see some interesting words such as ~booed, reflecting the fact that Mary was not booed, but that Cher was when the decision was made.

Finally the positive context analysis shows that the majority of tweets are about Mary 'going home' but also some positive sentiment such as: love, reliable and fabulous. It would not seem that there is great disagreement with the decision from the majority of the public.

This now leads us to wonder who will win? We will continue our analysis during the week to see what trends are forming from our tweet analysis.

X factor result show 6th December 2010

After last night's show there were huge numbers of tweets discussing the performances of the various contestants. We will present here a summary of our contextual analysis of all five contestants, and what this may mean for them in tonights results show.

To start with, we can look at how each contestant fared when analysed in context with winning words - i.e. which contestants, based on the twitter data, look most likely to win the public vote tonight?

Our analysis has the following ranking based solely on positive contextual analysis:

1. Cher
2. Rebecca
3. Matt
4. Mary
5. One direction

This result is particularly interesting.  One Direction have in the past always scored incredibly highly in our analysis, based on their demographic and how well it matches against the twitter demographic.  However, this time they are bottom alongside Mary.  It should be noted that Mary and One Direction scores are almost the same, with a clear gap to Matt and the others above them.

This would suggest that contrary to popular belief, that they may be in trouble in tonight's vote.  Mary also is in trouble looking at the scores, although she has been trending poorly for a number of weeks now.

We can also look at negative context, which means examining which contestants are talked about on Twitter as being who the public do not want to win.  This gives the following ranking:

1. Cher
2. Matt
3. Rebecca
4. One Direction
5. Mary

However, this ranking does not really tell the whole story.  The scores for each contestant are remarkably similar, and so there is not a clear difference between Cher ranked 1 and Mary ranked 5.  This is particularly worrying for Mary, since as we have seen previously she tends not to have as many tweets about her performance.  This would suggest that there is therefore a strong anti-Mary sentiment, when compared to Cher and the other performers.

Also, it would suggest that Cher is very much a contestant that you either love or hate, since she has finished top of both lists.  It is noticeable that Rebecca has moved down the negative context list, meaning that there are comparatively fewer people who dislike her performances when compared to the other contestants.

We can also look at examining the contextual analysis for individual contestants, such as Rebecca.  The figure below shows the analysis of the positive context for Rebecca.  Here we can see lots of positive context coming through, indicating that the majority of tweets were supportive of her performance.

As we can see, there are a number of positive words coming through such as: win, beautiful, amazing, best, grace.  If we look at the negative context for Rebecca then we can identify what issues people had with her performance.  The figure shows the negative context words closest to Rebecca.  The ~ symbol before each word indicates that to understand it, you should imagine a 'not' in front of each word.

Here we can see a number of words coming through, such as: ~dance.  This shows that there are a large number of tweets discussing the fact that Rebecca cannot dance, or does not dance, in her performances. Or perhaps, its because they think she's not suited to singing a dance song. We could create another context cloud for dance to reveal why and give more detail.  We can also see that not everyone likes her voice, as ~voice also comes up close in context.

We can undertake a similar analysis for Mary, where we can see the negative context analysis.

Here we can see that there is discussion about her not returning to Tesco as a checkout girl.  But we can also see that there is rather negative sentiment such as ~good, indicating that her performances were not good, and also ~win, indicating that there are a large number of tweets discussing the fact that they do not think that she can win the X Factor.

It will be extremely interesting to see the results of tonights show, in particular if One Direction do end up in the bottom two.  However, as always, we should take note of the fact that the twitter demographic does not necessarily match that of the voting public.  It is free to make a statement on twitter, but it costs money to actually make a vote.

We will analyse the results show tomorrow, to determine the public's view on who is actually removed from the show, and whether they agree with it.