EVOLUTION OF A MODEL TO DETERMINE UNSECURED TRANSACTIONS

Abstract

The widespread presence of fraudulent transactions in financial institutions is of significance in banking operations. Examples of financial instruments that are utilized include credit cards, smart cards, swipe cards, etc. These cards provide important information and enable small costs to be incurred by customers. These small amounts are removed from customer accounts. Banks need to discover the correctness of transactions, thus the introduction of the evaluation of models to determine unsecured transactions. The focus of this research is to contribute to the field of the application of machine learning to banking operations by introducing tools for predicting unsecured transactions in the banking sector. The research objectives include the examination of different methods utilized in machine learning for investigating unsecured transactions about the physical stealing of credit cards and the illegal collection of details on credit cards. To accomplish the aims of this research, information gathering is done using Kaggle. Kaggle is obtainable online. The major focus of this research is to examine cardholders' spending patterns. The method includes using a multilayer perceptron (MLP). This is utilized with training of 70% and testing of 30% subsets. The evaluation of the model is done using a confusion matrix technique. This research is implemented using the Python programming language. The model produces accuracy rates of 93% and 99% respectively.  This research can leverage achievements recorded to improve security concerns in financial institutions.