@takumamih On top of what @Reeza wrote: If you've got the Data Quality Server licensed then you can also create match codes using dqmatch() and then join over these match codes. http://support.sas.com/documentation/cdl/en/dqclref/70016/HTML/default/viewer.htm#p09nffezbjyj4on11oblz77aq1x6.htm   Based on the data you've posted you would still need to implement some data prep logic before feeding into dqmatch() as the strings look too different to each other as that I'd expect dqmatch() to return the same match codes for values you'd like to get matched.   What using dqmatch() would solve for you: 1. You're creating first these match codes on row level (that's the fuzzy part) but then join via an exact match over these match codes and though don't need anymore compare everything with everything. 2. You take advantage of a knowledge base (QKB) which comes already with a lot of pre-defined rules so you don't need to start from scratch. ... View more

Your code does not make sense. If you already have a dataset, you need no input statement, as the set statement does the read. If you want to read from an external file, you have to use an infile statement to open the file for INPUTting.   Post an example of your raw data file, as a properly created data step will very probably fix your problems in one sweep. ... View more