The Vesta dataset was released for use in the IEEE CIS Fraud Detection competition. It contains 590,540 card transactions, 20,663 of which are fraudulent (3.5%). Each transaction has 431 features (400 numerical, 31 categorical), along with the relative timestamp and a label of whether it was fraudulent or legitimate. For anonymization purposes, the names of the identity features have been masked, along with the names of the extra features engineered by Vesta.
Paper | Code | Results | Date | Stars |
---|