I am currently working on evaluating the result of my clustering algorithm against a predefined clustering "gold standard" data set.
According to my research the Fowlkes-Mallows Index should be a good tool to get a comparable evaluation score, but I am not sure about the correct application.
Currently I calculate the true positives, false positives and false negatives for each cluster, then sum them together for total values and add them in the formula.
where TP is the number of true positives, FP is the number of false positives, and FN is the number of false negatives.
Is this the correct application of the Algorithm? Or should I combine the results for multiple clusters seperatly? And if so, how?
This implementation of the Fowlkes-Mallows score in python might be helpful. Further information regarding the formula they used is provided here.