Credit Card Fraud Detection - Turkish Data

This dataset was provided by the Turkish company Yapi Kredi Teknoloji and contains a set of 62,380 transactions, of which 61,572 (98.70%) are genuine, and 808 (1.30%) are fraudulent. This imbalanced dataset has 26 variables: the transactions timestamps, the class label (1 for fraud and 0 otherwise), and 24 additional explaining variables. This dataset was also made available in two parts, being one of them for training/testing, with 31,190 transactions (30,400 genuine and 790 fraudulent) and the other, containing 31,190 observations (31,172 genuine e 18 fraudulent) for validation. Despite the transactions represented in this dataset being actual, the variables are numeric, resulting from preprocessing by Principal Component Analysis (PCA) [43], and do not have a description due to confidentiality concerns.