3 ms·
Ugh. At a past job I was tasked with matching records, based on street addresses and business names, from various systems. Levenshtein Distance was pretty instr
by stanmancan 7y ago
Ugh. At a past job I was tasked with matching records, based on street addresses and business names, from various systems. Levenshtein Distance was pretty instrumental in helping find potential matches once we had cleaned up the address format and business name as much as possible.
- tongo 7y agoI did the same thing. I was merging our billing db with that of a business we had just bought. Due to the nature of our business, customers could exist in both. So I had a routine that normalized the address, as best I could anyway, didn't help that it was all just one varchar field. Then implimented Levenshtein Distance, messed with the weighting a bit to fit our particular data and away it went. Saved a bunch of headaches. It wasn't perfect, but it was better than hand matching a couple thousand accounts.