Since Italy is now out of the soccer World cup 2010, I can talk about data mining in another domain: auto insurances.

A few weeks ago I found this interesting article on auto insurance policies for men and women: http://articles.moneycentral.msn.com/Insurance/InsureYourCar/worse-drivers-males-or-females.aspx.

Insurance companies are known to be profit driven. If they make a discount to some insured people group, usually they have their good reasons.
In this article a few statistics are shown about traffic violations, car crashes, and D.U.I. for women and for men. From those statistics women seem to be less reckless drivers than men.

Somebody could object that you can show anything with a few statsitics, as long as your statistical sample is biased, your calculations are biased and so on. However, the article also shows that the majority of insurance companies in the USA have different policies for men and women... and usually the car insurance policies for women are cheaper.
This should indirectly confirm the statistics, based on the assumption that insurance companies are profit driven.

However, at least some of those gender differences seem to be due to life style. For example, the percent of D.U.I. cases has grown for women from 1998 to 2007 and ddecreased for men in the same time span.

 Since we are getting closer and closer to the next soccer world cup (2010), I thought I could analyze the performances of the soccer teams during the last world cup (2006). The data for the last world cup is available on the FIFA web site (http://www.fifa.com/worldcup/matches).

Data analysis is definitely not going to tell me who the next winner is going to be, because I miss the data for the last 4 years for all participant teams.
However, I can try to quantify the teams performances for the last world cup.
Let's call it a retrospective study.

From the FIFA web site with the match results I have built a file "soccer 2006.csv" with:
      - number of games played by each team (I did not include the final game for 3rd and 4th place)
      - number of goals scored
      - average number of goals scored per game
      - number of goals taken
      - average number of goals taken per game
      - a fit measure defined as:      (nr of scored goals - nr of taken goals)/nr of games

I apologize here if my english soccer terminology is not perfect. I must confess I am not an expert of the field.

Total number of scored goals
If you process the file and sort the data by "#of goals", you will see that Italy has the highest number of goals (12), followed by Argentina and Germany (11), Brazil (10), France and Spain (9), England and Portugal (6) , Australia, Cote d'Ivoire, Ecuador, Mexico and Ukraine (5). 

Average number of scored goals per game
However, if you sort the data by number of average scored goals per game, Spain moves into the first position (2.25), followed by Argentina (2.20), Brazil (2.0), Germany (1.83), Italy (1.71), Cote d'Ivoire (1.67), France (1.29), Australia (1.25), Ecuador (1.25), Mexico (1.25) England (1.20).

Total number of taken goals
But soccer is not only about scoring goals. It is also about not taking any!
If we check the total number of taken goals, we find the following ranking:
- Switzerland (0)
- Portugal , Paraguay, Brazil, Netherlands, Italy, England, Angola (2)
- France, Croatia, Argentina (3)
- Spain, Trinidad and Tobago, Poland, Korea Rep, Czech Rep., Sweden, Ecuador (4)
- Mexico, Germany, Togo (5)

Average number of taken goals per game
Let's check now whether the taken goals were distributed on the same number of games.
- Switzerland (0)
- Italy (0.23)
- Portugal (0.33)
- Brazil, England, France  (0.4)
- Netherlands (0.5)
- Argentina (0.6)
- Angola, Paraguay (0.7)
- Germany (0.8)
- Spain, Mexico (1)

The fit measure
Finally, let's check the fit measure (nr of scored goals - nr of taken goals) /nr of games.
Of course, here we can object that not taking goals is more important than scoring them or viceversa. Anyway, with this fit measure we get the follwing ranking:
- Argentina, Brazil (1.6)
- Italy (1.4)
- Spain (1.25)
- Germany, Switzerland (1)
- France (0.86)
- England (0.8)
- Portugal (0.7)
- Ecuador, Netherlands (0.25)
- Mexico (0)

These are just a few numbers from the world cup 2006 to play with. I do not think that we can infer anything from them to predict who is going to win the world cup in 2010. 4 years have passed, the teams are not the same anymore and a lot of other things have happened. But still it is interesting to be able to quantify the teams' performances.
I claculated the same numbers for the world cup 2002.
I will do the same for this soccer world cup 2010 once it is over.

Just contact me, if you want the full presentation and the other numbers.