Hi HB, I did these process for acqusition campaigns for a year; matched third party several million records with existing customers database; the way I did kind of like baysien rules hopfully these help you or any other people who have the same issues; e.g. column required: name / address/ postal code; 1. match all records by postal code level (really depends on database size, if you have more than million records it will take really long time for entire fuzzy matching process), 2. most of address read as suit # - street number street name or street number street name suit #; try to scan the first / second word (street number) to reduce the records numbers; (Po Box may run seperately) 3. scan two or three street name words use function compged to fuzzy match; then you will have matched at building level; 4. if you have name almost do the same thing as step3; I pretty sure you may have a really good match rate; PS. necessary data cleanning is required before the whole process; e.g six -> 6, highway -> hwy Hopefully these work for u; Good luck Hao
... View more
