I was reading a paper the other day on Machine Learning framework for Sports Performance by Rory P. Bunker and Fadi Thabtah, here for full paper. In it, the authors said:

There is unlikely to be a great degree of imbalance in the class values for the dataset, although given the commonly observed home advantage phenomenon, one is likely to see a slight skew in favor of home wins. In this case, classification accuracy is a reasonable measure of evaluation. In cases where the data is highly imbalanced, ROC curve evaluation may be more appropriate.

Now I am still fairly new to some of the terminology, but are they arguing, that when data is imbalanced, to test the performance we should use the ROC value or the Area Under the ROC Curve, AUC, value? I interpreted it as AUC because they state evaluating the curve may be more appropriate, but I may have misinterpreted the context.

Secondly, within my dataset, I have 45% home wins, 30% away wins, and the last 25% as draws. Would others believe that this is highly imbalanced?


Your Answer

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

Browse other questions tagged or ask your own question.