Understanding Nature of Problem in case where test data is not labeled

I understand that Supervised Learning are those problems where the data are labels and unsupervised Learning are those problems where data are not labels. I am solving a classification problem where dataset is like base on some historical performance of teams and I gonna predict which team will win or loss.
But I have data set for training and testing where training is labeled but test set are not labeled. Now if I applied a classification algorithms on this type of problem How can I evaluated my model in this case.

Hey @Abbas_Malik,
As per the definition, training, dev and test sets in a supervised setting are supposed to have labels. If your “test” set doesn’t have labels, then you can’t use it for evaluating your model’s performance, unless and until you label it, for instance manually. If you can’t do that, then simply split your training set into training, dev and test sets and then you can evaluate your model’s performance.

Another possibility is that you have a “target” model based on which you want your model to learn. In this case, you can simply use the target model to predict the labels for the test set, and then use the test set to evaluate your model’s performance. I hope this helps.