The project intends to predict the European soccer match outcome from 2008 and 2016. The features they are interested in studying include but not limited to players' rating, geographical origin and players' lineup positions.
After evaluating this report, I have summarized the following three things I like and three other things that could be improved.
Pros:
1) I like the way they cleaned their data. They first examined the overall data, dropping all empty or NA data. They have about 20, 000 data left.
2) I like their idea to first run linear regressions knowing that classification model such as random forest might work better, just to get a sense of what linear regressions tell them about their problem.
3) In their final section about future plans, I like their ideas about studying more features to better predict the outcome. For example, the lineup position of the initial players might be worthwhile investigating.
Cons:
1) I have some concerns about their project objective. If their only purpose is to predict the match outcome in the past years, then the project is bit worthless. However, I see their potential in using this model to predict further match outcomes. I have not seen many details about how to use their model to predict future soccer outcomes yet. Thus, it would be nice if they can discuss then in their final report.
2) in section 2 about using Random Forest, they mentioned that their prediction result is better than 10 other bookmakers's predictions. I did not see any quantitive result proving it.
3) Still in section 2, their picture showing their roots and nodes of random forest is helpful. However, It would be nice if they can label them to show what the "branches" actually mean.
The project intends to predict the European soccer match outcome from 2008 and 2016. The features they are interested in studying include but not limited to players' rating, geographical origin and players' lineup positions.
After evaluating this report, I have summarized the following three things I like and three other things that could be improved.
Pros:
1) I like the way they cleaned their data. They first examined the overall data, dropping all empty or NA data. They have about 20, 000 data left. 2) I like their idea to first run linear regressions knowing that classification model such as random forest might work better, just to get a sense of what linear regressions tell them about their problem. 3) In their final section about future plans, I like their ideas about studying more features to better predict the outcome. For example, the lineup position of the initial players might be worthwhile investigating.
Cons: 1) I have some concerns about their project objective. If their only purpose is to predict the match outcome in the past years, then the project is bit worthless. However, I see their potential in using this model to predict further match outcomes. I have not seen many details about how to use their model to predict future soccer outcomes yet. Thus, it would be nice if they can discuss then in their final report. 2) in section 2 about using Random Forest, they mentioned that their prediction result is better than 10 other bookmakers's predictions. I did not see any quantitive result proving it. 3) Still in section 2, their picture showing their roots and nodes of random forest is helpful. However, It would be nice if they can label them to show what the "branches" actually mean.