I recently got into a topic regarding fraudulent transactions. I am relatively new to data mining and just looking for some input for my case here.

I started with a cluster analysis / anomaly detection and got some good results in the first step. Now I want to create some more analysis, but I am a little bit confused about which data mining algorithms and methods fits best in my case.

what do you think about pattern mining, for example with apriori or something? patterns that appear less often are probably an "outlier" too and could be used for further mining in the data.

its not just creditcard transactional data, its data from a point of sale.

  • 1
    $\begingroup$ from experience i have found benfords law to be very useful. $\endgroup$ – user73824 Apr 21 '15 at 9:31

I think we could use 5 method:

  1. find outliers objects
  2. find anomaly objects with clustering
  3. use business rule for find
  4. use previous fraud to find new similar fraud
  5. find change in pattern of transaction of each objects

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.