Is there a way to "measure" some kind of relation between a variable (in a power law distribution) to the data itself.
For example, in normal distribution, I can use mean and std to get a sense of how related a certain variable is to the normal distributed data,
I would like to do the same, but my data is distributed according to power law.
The final idea is, to create some kind of Hash functions that will map a variable according to its "relation" to the data, such that similar variables (i.e. that have same/similar "attributes") will fall in the same bucket.
Hope I made myself clear...