My reply is not intended to be a comprehensive overview of fuzzy matching. There are lots of SAS samples on the internet. Soundex was designed by genealogists to deal with differences in spelling of names. It doesn't handle numbers well (for example, 1 is not interpreted as ONE, both 112 and 1122 are simply treated as numbers so the soundex function will miss the similarity entirely). It doesn't expand abbreviations (AVE is not expanded to AVENUE, FT ST JOHN is not interpreted as FORT SAINT JOHN), or vice versa. So, soundex will work reasonably well on some variables (e.g. City or State), but less well on variables such as Street. However, some data cleansing may be required prior to attempting a match (e.g. ensuring State names are all abbreviated or vice versa). There are other SAS Functions that can help with fuzzy matching, and that may be superior to soundex depending on the situation. (Example: http://blogs.sas.com/content/sgf/2015/01/27/how-to-perform-a-fuzzy-match-using-sas-functions/)
... View more
