Online Transaction Fraud Detection: Exploring the Hybrid SSA-TCN-BiGRU Approach

The ease of online payment has lowered the geographical limits for shopping, opening up several new options for e-commerce in the past decade. Online shopping may be a great way to save money, but it has also made it easier for con artists to acquire sensitive information like credit card numbers. A new approach is proposed in this study for the purpose of detecting fraudulent online credit card transactions. The proposed approach consists of preprocessing, feature extraction and training the model. Managing Missing Data and Handling Outliers are two aspects of data preparation. PCA is used in feature extraction to reduce the number of correlated variables in a collection to a smaller number of uncorrelated variables. We used the SSA-TCN-BiGRU to increase the accuracy of the model training procedure. The suggested method outperforms the methods of its competitors, including SSA and TCN, with an accuracy of 97.50%.