03-22-2015 05:20 PM
I work in credit card fraud and recently got access to SAS Enterprise Miner and SAS Enterprise Guide. I have a fairly good knowledge of SQL so I've been able to use SQL in Enterprise Guide to work with datasources I wasn't previously able to work with using SQL which has been great. What I'd really like to do though is figure out how to use link analysis in Enterprise Miner to identify compromised merchants. All of my transaction information is stored in one table. This includes the account number, merchant name, date, time, whether or not a transaction is fraud etc. I know I need to pull out the transaction data for just the cards that have fraud and then set that up as transaction data but I've read through tons of pages of tutorials and can't figure out how to do this. It took me forever just to figure out how to put my LIBNAME statement in the project start code so that I could add my data source as input data.
None of the tutorial information I have found seems to explain how to do really basic things or if it does it's specific to a case study and doesn't give a general explanation. Can anyone point me towards information that will help me accomplish what I'm wanting to accomplish? I do have some programming experience so I feel like I can figure this stuff out if I can find the information I'm looking for. I'd be extremely grateful for any help you can provide. Thanks!
03-22-2015 07:26 PM
03-23-2015 12:37 AM
I don't have a ton of experience with EM, only a project or two, but I found that preparing the data via code first and then bringing it into EM was what worked best. I also did some of my exploratory analysis outside of EM until I needed tools that were in EM specifically.
For learning EG, I'm not sure what resources there are, but for learning by code SAS Analytics U has a free e-learning course available.
03-23-2015 09:34 AM
Yes, if you are comfortable with EG, by all means prepare your data there and import it into EM using an input data source node. Be sure to set role=transaction.
This paper that gives some background on Link Analysis as well as some examples:
If that's too much to digest at this point, you could start by doing example #2 "Examining Web Log Data" in Enterprise Miner Help for the Link Analysis node.
Hope this helps.