Track A dataset - errors or genuine features?

When looking at the data provided, I noticed that the swift dataset contains senders/receivers that are not found in the bank dataset. There are also OrderingAccount and BeneficiaryAccount that are not found in the bank dataset.

How should team interpret this? Is this:

  • an unintended error in the synthetic data generation
  • an intended error in the synthetic data generation (i.e. this reflects a real-life error the datasets might contain)
  • a relevant signal for detecting anomalous transactions - I note that every transaction associated with a bank or account that is not found in the bank dataset is flagged as anomalous
2 Likes

Hi @HectorP, please treat this part of the dataset as is. It is not an unintended issue with the data.