%LET BAD_PART=DUPLICATE|DO NOT USE|DONOT USE|FILE|RECORD|DO NOT ISSUE|DONT USE; DATA test1a; SET test1; ARRAY ORIGINAL [*] $ 40 C_LN C_FN; ARRAY NAMES[*] $ 40 LN FN ; DO I = 1 to 2; NAMES [I]= ORIGINAL [I]; END; DO I = 1 to 2; /* turn anything that is NOT a letter into space*/ NAMES[I] = prxchange("s/[^A-Z]/ /",-1,NAMES[I]); NAMES[I] = prxchange("s/"&BAD_PART"/ /",-1, left(trim(NAMES[I]))); NAMES[I] = STRIP(NAMES[I]);/*removing all front and back space*/ NAMES[I] = COMPBL(NAMES[I]); /*Change multiple blanks into one*/; END; DROP I; RUN;   It won't run. Can someone help me? Thank you!  @maguiremq ... View more

I have two datasets that I need to match based on names. However, the names can be really messy. I have two main issues. First, for example,  Dataset A                 Dataset B Name                        Name De la Rosa               Maria Rosa De la Maria              Rosa Maria de la Rosa     Maria de Rosa Maria Laura             Maria Maria Gabriel Rosa   Gabriel   The standard that I consider true match will be De la Rosa and Rosa, De la Rosa and Maria Rosa, Maria de la Rosa and Maria de Rosa, and Maria Laura and Maria. And Maria Gabriel Rosa would be a match with either Maria, Gabriel or Rosa. But Maria Laura and Maria Rosa will not be considered as a match.   So basically if a name has more than one part with meanings, eg. Maria Rosa (de la is not considered as meaningful), it would be a match with names of either of the individual part, eg. Maria. But it would not be a match with another name with two meaningful parts, like Maria Laura.    What I think may help is to get ride of strings like "de", " de la" (I have a bunch of others with no meanings in Hispanic names) and create new names like: Dataset A                                  Dataset B name                    new_name           name                 new_name De la Rosa             Rosa                    Maria Rosa       Maria Rosa De la Maria            Maria                  Rosa                   Rosa Maria de la Rosa    Maria Rosa        Maria de Rosa    Maria Rosa Maria Laura            Maria Laura        Maria                  Maria Maria Gabriel Rosa Maria Gabriel Rosa   Gabriel        Gabriel   I think i can use prxchange to do that. However, I still don't know how I can make sure all the matched are found, considering there are names with equal and more than two meaningful parts (I have names up to four meaningful parts).   The second issue is that there are a bunch of noises in the names, eg. Do not use, Duplicate, Gone, no existing file. They do not always stand alone. It can be "Maria Do not use", "Jose (Gone)", [OLD]John. I'm thinking to use macro to get rid of them and add newly found ones to the macro for future years of data match. I wrote this: %LET NOISE = "DO NOT USE","GONE", "DUPLICATE"; If find (name, "NOISE")>0, then do; new_name=prxchange("s/(&NOISE)"/ /,name);   On top of that is that I have really large files with tens of thousands of records. I'm just really stuck.  Thank you for any suggestion.   ... View more