There are certain input-output processes that display a "20-80 relation". That is, given Y depending on X, there are subsets of X consisting 20% of the input set that generate 80% of the output. For example, 20% of clients may generate 80% of revenues (or complaints), 20% of product flaws generate 80% of complaints, etc.
Are there known methods to see if a given data set satisfies this 20-80 principle?
