Hi! Sorry if this has already been posted but cannot seem to find a previous post on this.    I have a variable in a large data set (~50,000 obs so too many to manually correct) that can take on 1 of 3 possible structures: 1. name (example: Hartsfield airport)  2. xxxx - name (example: 12345 - Hartsfield airport)  3. name - xxxx (example: Hartsfield airport - 12345) Here is some code to create a sample in SAS.   data trial; input have $; datalines; "12345 - hartsfield airport" "hartsfield airport - 12345" "hartsfield airport" ;   The strings are of many differing lengths and the numbers can be any combo and length so my sample may be oversimplified. I am wanting just the name out of the variable's string. There is no rhyme or reason for there being numbers and a dash before or after the name so I cannot subset the datasets based on another variable and use a simple strip(scan(have, 1, "-"))  or strip(scan(have, -1,  "-")) to extract just the name. I really wish it was that simple.  Is there a way to modify the strip and scan functions or another function I should use to extract name?    I have thought about using something like the FIND, INDEX, or ANYDIGIT function to do a subset and then apply the strip/scan functions but because the numbers can take on any value and the strings have varying lengths, I don't think those will work. Any advice is appreciated. Thanks!      ... View more

Hello,  I am matching 2 relatively large datasets (both over 30,000 observations) based on three variables: facility name, facility state, and facility zip. The facility names are not always a perfect match between the 2 sets. For example: Set 1 Saint Joseph's Care at Boardwalk and Center for Health Hospital at Mercy Set 2 St Joseph's Hospital Care at Boardwalk Mercy   As you can see, the names may have different spellings (Saint vs St) or include/exclude certain words (Care at Boardwalk and Center for Health vs Care at Boardwalk, Hospital at Mercy vs Mercy). There is no particular rhyme or reason for the differences in name so using a substring function to extract certain words would not be ideal. Using a simple proc sort and match-merge does match the ones with exact matches but I still have 20,000 observations that are not matching. Are there any functions that may be particularly useful for this kind of problem? I would greatly appreciate any help in getting SAS to do the matching for me.  ... View more