Your project tries to use European soccer match information to predict outcome of the match and you have a comprehensive explanation of the dataset as well as a preliminary analysis of the data to provide guideline for further analysis. I like the idea of using visual graphics to present the data, which gives viewer a rough idea of what the data is about and can extract information directly without going though the explanation. You used simple and neat way to process data. By adding some simple features to the input and output space you optimized the dataset for regression analysis.
For first attempts at the problem you used 2 methods: linear regression and random forest. As you pointed out linear regression is not optimized for classification problem and as a result the coefficients are not suggestive. As for random forest model, in comparison to the linear regression model, gives much cleaner and more accurate output.
I agree that you need to have more features in the input space to enhance accuracy of the model and enhance robustness. In addition, I think you would benefit from testing some other models on the data and select the best one through cross validation.
Your project tries to use European soccer match information to predict outcome of the match and you have a comprehensive explanation of the dataset as well as a preliminary analysis of the data to provide guideline for further analysis. I like the idea of using visual graphics to present the data, which gives viewer a rough idea of what the data is about and can extract information directly without going though the explanation. You used simple and neat way to process data. By adding some simple features to the input and output space you optimized the dataset for regression analysis. For first attempts at the problem you used 2 methods: linear regression and random forest. As you pointed out linear regression is not optimized for classification problem and as a result the coefficients are not suggestive. As for random forest model, in comparison to the linear regression model, gives much cleaner and more accurate output. I agree that you need to have more features in the input space to enhance accuracy of the model and enhance robustness. In addition, I think you would benefit from testing some other models on the data and select the best one through cross validation.