I am working on a determining why certain employees cause errors in a company's process and why others do not. I have the employee information, information about the errors they have made and the teams that the employees are in.

All the employees in all the teams have caused errors at one point or another. What i want to find out is if there are certain characteristcs about certain employees that causes them to make certain types or number of errors. The sample size is a few thousand over a six months period.

What do you guys think is the best approach, i.e. clustering method/ general data mining method?

  • 1
    $\begingroup$ Can you clarify how an error is measured or quantified? $\endgroup$
    – Maurits
    Feb 23, 2012 at 12:06
  • $\begingroup$ Errors can be anything from wrong spelling of a name to wrong account number setups. All errors are weighted equally. So errors are all integer type. If you make both of the above errors you will have made 2 errors. Separate variables record information such as time of the error, type of error, who caused the error, who detected the error, difference between when the error was made and when it was rectified. I also have information on each employee that I can wash against the data I.e. hr information. $\endgroup$
    – DataDancer
    Feb 23, 2012 at 19:49

1 Answer 1


The best approach seems to be using Bayesian networks, which are used for just that purpose. Here's a free tool for automating the process.

Depending on how much effort you're willing to invest, you can go all the way to causal analysis and intervention calculus, which are the natural next step.

  • $\begingroup$ Thanks! i will give these a go and let you know how it goes $\endgroup$
    – DataDancer
    Feb 28, 2012 at 4:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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