I have two unlabelled data sets A, B for student marks, the two data have the same number of objects or records and the same features or attributes, marks of (Maths, Physics Chemistry English) but for two years. I applied Fuzzy C-means to each one of them and got 4 clusters for each one, as we know the unlabelled data will be labelled data after the clustering process. I am confusing about measuring the similarity between the results, most the external measures are applicable for data with truth classes, I mean the comparative between truth and predict classes, what about comparative between predict and predict?
Any suggestions may help me.
thank you in advance.