I have an (unbalanced) panel dataset with 20 countries, 57 years, and 8 variables, and I would like to cluster the countries according to their dynamic trend in these variables (whether using kmeans or hierarchical clustering does not matter).
I have seen this question and this paper which try to assess the issue but I would like to find a method based on a distance measure on which to implement the usual "cross-sectional" clustering algorithms.

In other words, I am looking for a dissimilarity measure to be applied to multivariate data, describing the diversity between the shape of individuals' time trend in the variables of interest.

Any suggestion on panel clustering in general are also welcome, thanks!



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.