Ordinal Regression/Classification is a classification where the labels have orders (https://en.wikipedia.org/wiki/Ordinal_regression)
Can you comment what are pros and cons if someone wants to solve a regression problem by ordinal regression? Why someone should or shouldn't do that?
Why I am asking:
I work on a regression problem with binned inputs. So, the cardinality of input values is a number like N and thus the cardinality of all output is n << N.