I am working on developing an predictive system which is capable of predicting the match winner in a football game.

I have 2 datasets,

  1. Players related data- contains 88 features.

  2. Match related data- contains 10 features.

As of now the solution I have thought is to develop a player rating based on different feature values, aggregate the rating based on team and add this as one of the feature to match related data and apply classification algorithm on match data.

As an initial stage, what would be the efficient way to provide ratings to all the players ?

Is there any other better solution so that I can also include players quality in predicting the match outcome rather than just depending on precious match results.?

Thanks in advance


With a problem this big, it's hard to guess what methods would work well in advance of actually working with the data. The first thing you should do is look for previous research on predicting match outcomes in your game (which could be association football or gridiron football; you didn't say). Reading more widely about predictive data analysis in sports could also be helpful.

If the player composition of teams doesn't change too much in your dataset, it's hard to see how player data would help you predict match outcomes. So, it would make sense to ignore the player data entirely and conduct only team-level analyses.

  • $\begingroup$ Thanks a lot for your reply. I am working on association football, Bundesliga is my particular point of interest. Say, If i follow my solution which i have mentioned in my post, how do i proceed to develop player rating for individual players ? $\endgroup$ – Abhi Nov 7 '16 at 22:21
  • $\begingroup$ @Abhi I'm sorry, I don't know what I could say that would be helpful without knowing much more about the data and the problem. You might consider hiring a consultant. $\endgroup$ – Kodiologist Nov 7 '16 at 23:06
  • 1
    $\begingroup$ Will it be helpful if I can add the sample of the data? Also, I can elaborate the question in details if it might help you to suggest solution $\endgroup$ – Abhi Nov 7 '16 at 23:09
  • $\begingroup$ @Abhi I'm not sure I have the right skills for this, but editing the question to add details might help somebody else help you. $\endgroup$ – Kodiologist Nov 8 '16 at 3:16

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.