Fashion Retail Digital Payment Fraud

Dataset Description

The dataset comprises 1,000 individual transaction records, with each record representing a single digital payment transaction completed by customers within the fashion retail sector. The data is organised in a structured tabular format, where each row corresponds to one transaction and each column represents a specific attribute associated with that transaction. The dataset has been structured to support analytical examination of transactional behaviour and fraud-related patterns in digital payment environments.

A binary target variable is included to indicate the outcome of each transaction, allowing the dataset to be used for classification-based analysis. Alongside the target variable, the dataset incorporates a wide range of explanatory variables capturing demographic information, purchasing behaviour, payment characteristics, account activity, device usage, fulfilment details, and historical dispute behaviour. This multidimensional structure enables comprehensive representation of digital payment activity at the transaction level.

Target Variable

The dataset includes one dependent variable:

is_fraud

This variable indicates whether a transaction is fraudulent or legitimate. It is encoded as a binary value, where 0 denotes a legitimate transaction and 1 denotes a fraudulent transaction. The presence of this variable enables supervised classification by allowing models to learn patterns associated with fraudulent behaviour in comparison to normal transaction activity. The distribution of this variable reflects a typical fraud detection scenario in which fraudulent transactions occur less frequently than legitimate ones.