Detect Bid Rigging in dataset, using Auto Model

Ric1

Hi, I would do a bid rigging analysis with RapidMiner, using Auto Model.
I am not an expert, so I need support from all of you.

So, in your opinion to find some bid rigging action in my dataset, what type of analysis do I have to do: Predict, Cluster or Outliners?

I have a dataset with attributes like: tender results (a flag awarded / not awarded), winner name company, winner bid amount, and other attributes; but I don't have an attribute that contains all the not winner offers of the participant, for each row surely.

So what would you do in my shoes?

Thank you so much for the future answer.

Best regards.
Ric

rfuentealba

Hmmm, this bothers me.

I have a dataset with attributes like: tender results (a flag awarded / not awarded), winner name company, winner bid amount, and other attributes; but I don't have an attribute that contains all the not winner offers of the participant, for each row surely.

Anyway, rigging bids is a kind of fraud. If it's subtly done (I'm not a fraudster, but I research lots of those) it would be kind of an anomaly or outlier detection. On the other hand, you should do your statistics, check the probabilities of a certain bid to be rigged (depending on the parameters) and use classification algorithms such as decision trees or probably regressions (pretty much in the same vein as credit risk scoring). It depends on the kind of data you have, how much comparison data you have and so on.

Ric1

For example, if my dataset is form by these attributes:

Lot ID (type string of number),

Lot Name (type string),

Participant Company Name (type string),

Participant Result (string with 2 possible value: winner / loser),

Award Date (date);

what attribute do you choose for a Bid Rigging analysis, using Auto Model and in particular the Predict section of it?

What is the best attribute for the Predict section, to find an eventual event of Bid Rigging in the dataset?