Chris B on 24 Feb 2023 23:36:53
It would be awesome if there was an option to change the algorithm used for fuzzy matching.
At a minimum I think it would be obvious to provide a version of the matching functions that take a parameter `matchFunction(str,str)=>(num between [0,1])`. Then if our friendly neighborhood Microsoft devs are feeling generous, the library could provide also provide functions Match.LevenshteinEditD (what I really want), Match.JaccardIndex (the current option), and whatever other similarity functions, all as first class library functions.
This change could even allow for extending fuzzy matching to additional kinds of data by providing matchers that work on them in the future (I'm picturing bitmap data, or audio/signal clips).
None of this would break current code, but it would expand the utility of fuzzy matching in Power Query significantly. I am aware that one can rig something together with the comparer version of Table.Group and/or other functions--along with liberal application of custom M code--but this would enable non-M savvy users to do so as well.