I would like to perform a cluster analysis but I’m not sure which is the best algorithm to apply to my data.
My dataset is made of 200 cases (but the sample size can be enlarged). For each case, I have 15 different variables that are measured in two subsequent moments. All of the 15 measurements have an identical ordinal likert scale ranging 0-4. Treating them as an interval scale might not be a correct assumption.
I don’t have any a priori hypothesis regarding the number of clusters, not even the shape of clusters. I would like to be as much flexible as possibile and a method that take the temporal nature of the data would be a desired plus!
What do you suggest?
Thank you very much!