I am trying to implement a visual representation of a number strings to make it easier for the user to identify similar values. The first part of this is generating a fixed length hash of the data – however I am not aware of an appropriate hash algorithm.

All my Google searches so far have returned endless references to cryptographic hashes; where a small change in the input produces a big change in the output – exactly the opposite of what I want to achieve here (indeed a surprising number of commentators think the only type of hash is a cryptographic hash).

Are there hash algorithms available which will do this?

(I understand that a hash behaving as I describe would have high collision probabilities – this is only a minor concern)