A MODEL FOR CREDIT CARD FRAUD DETECTION

ABSTRACT

Traditional methods of making payments by hand have been rendered obsolete as a result of rapid technological advancements in electronic commerce and widespread adoption of internet-based financial transactions. More and more people are turning to their credit cards while making payments and buying things from online stores. The anonymous nature of the internet and the high volume of credit card transactions has brought about a surge in the number of people who conduct fraud, as well as chances to perpetrate crimes such as phishing, illicit fund transfer, soliciting of monies online, and credit card information theft. This has led to a loss of billions of Naira each year, in addition to bringing untold hardship to members of the general population who were unaware of the danger. The conventional method of identification and authentication is based on possession of a PIN and password, as well as third party authentication using one time pin. However, these systems are not reliable because fraudsters are adaptable and, given enough time, will typically find ways to circumvent such measures. In recent times, a number of preventive and detection systems have been proposed and developed to enhance or mitigate credit card fraud. These systems include the conventional method of identification and authentication. As a consequence of this, this system, which will detect fraudulent activity involving credit card transactions in real time, was conceived of and developed. The purpose of the proposed system is to identify irregular or fraudulent financial dealings through the development of a machine learning model utilizing the Random forest classifier Algorithm. A model that can identify unusual financial dealings was trained and tested with the help of a dataset that contains bank transactions and information about people who hold credit cards. In the context of this project, the system was developed using the computer language Python, and the coding will take place in a Jupyter environment. The proposed system efficiently predicted with 99% accuracy.