• Hi All

    Please note that at the Chandoo.org Forums there is Zero Tolerance to Spam

    Post Spam and you Will Be Deleted as a User

    Hui...

  • When starting a new post, to receive a quicker and more targeted answer, Please include a sample file in the initial post.

Accolade Fraud Identification

M

MBB

New Member
Hello,
I run a call centre employee campaign in my company that tracks accolades different call centre employees get from their fellow colleagues for meeting their KPIs. I have downloaded a spreadsheet from the accolade system that shows 40000 (rows) accolades with the following column headers:

1. Accolade Receiver Employee ID (Col A) e.g. 112234
2. Accolade Giver Employee ID (Col B) e.g. 212345

In our company, anyone can give an accolade to anyone and employee bonusses also depend on these accolades. Management wants to identify cases of fraud where employees may have just exchanged more than 3 accolades between themselves to improve their KPIs. Can anyone help me what is the best way to do it?
In other words, I want to flag records in Col A if

1. Employee has received 3 or more accolades from the same employee i.e. values in Col A & Col B appear in the same sequence 3 times or more
Col A Col B
112234 212345
112234 212345
112234 212345
Ideally I want to pull these records in another spreadsheet with a title called - Potential Favorism

2. Employees have exchanged 3 accolades between themselves i.e. employee A gives 3 accolades to employee B and then receives 3 accolades from employee B in return
Ideally I want to pull these records in another spreadsheet with a title called - Potential Fraud Exchange

Can anyone tell me what would be the best way to do that?
 
OK - in the absence of an example file, I've created some dummy data and identified a couple of ways you might identify dubious combinations ....

The Table combines the Emp IDs of A & B, and reverses them - then it looks for the number of times one combination appears in another - mutual adoration of greater than 3 is filtered in the Pivot table.

Sherlock Holmes would have had a better way, but you will be able to identify any nefarious characters using this .... ;)
 

Attachments

  • Hemp Praise.xlsx
    16 KB · Views: 11
David Evans said:
OK - in the absence of an example file, I've created some dummy data and identified a couple of ways you might identify dubious combinations ....

The Table combines the Emp IDs of A & B, and reverses them - then it looks for the number of times one combination appears in another - mutual adoration of greater than 3 is filtered in the Pivot table.

Sherlock Holmes would have had a better way, but you will be able to identify any nefarious characters using this .... ;)
Click to expand...

Thank you very much. This is exactly what I wanted to do. Really appreciate it.
 
@MBB Interesting problem. I am afraid it is far more complicated than seeing
A->B , B->A accolades. What about circle jerks? (ie A praises B, B praises C, C praises A or A praises B & C and B&C praise D, D praises A or any of the other infinite variations?)

If you are serious about detecting all circular relationships, start here:

https://stackoverflow.com/questions/546655/finding-all-cycles-in-a-directed-graph

and implement this - http://www.cs.tufts.edu/comp/150GA/homeworks/hw1/Johnson 75.PDF

Or, talk to your employees and encourage them work with integrity and honesty :)
 
I am afraid it is far more complicated than seeing
A->B , B->A accolades.
Click to expand...
Yes, should have read the question in it's entirety and not just what's required output from given data. :p

I personally prefer to leave peer accolade out of performance evaluation, and base evaluation solely on hard stats. Or use 360 review instead of peer accolade. Since, 360 review is based on review from multiple layer within an organization (boss/managers, peers, subordinates etc) and is usually done anonymously, it gives better understanding of an employee's performance within the organization.

Peer accolade can be used for more fluffy HR program ;)
 
You must log in or register to reply here.
Back
Top