Hi all, I am working as a dba in a small bank and I got an assignment to develop an in-house tool for detecting fraud scenarios. I know the technical part of manipulating multiple data sources and mining data, but do you happen to know what tests/scenarios the banks use to detect fraud across all types of transactions (e.g. deposits, cards, online banking, employee fraud, wire and ACH etc). Have any of you coded such tests/scenarios? Thanks.
I would start researching "electronic bank fraud detection". Specifically look for banks, security vendors, or other industry entities that may offer training on such technologies. I don't specifically know who would be offering it, but I'm sure that it exists somewhere. Detecting fraud through data mining on any scale is well beyond the scope of most programmers. We're talking complex calculus, statistics and linear algebra. Just understanding any of this is difficult, and putting it into practice is exponentially harder. Here's a good document relating to credit card fraud: http://cs.fit.edu/~pkc/papers/ieee-is99.pdf The references in that article may be a good starting point as well. Each type of fraud is going to require a different method in trying to identify it. I would ask the bank what type of fraud are the most common and/or the most damaging and start there.
Thanks for the advice. This assignment was given to me with a tight deadline, I thougt that someone might have already implemented such types of fraud scenarios.
I'll be sure to post if I come across anything. This sort of information is usually well guarded. You might try to find out if the bank has a relationship with other banks. Contacting their tech departments may be a good way to get something.
im not sure even if someone knew on here youd want there help.... lil bit of extra code they add in and boom there goes millions along with your job