CREDIT CARD FRAUD DETECTION USING ADABOOST AND MAJORITY VOTING

Abstract

Credit card fraud affects the financial services industry greatly. Credit card fraud costs American
businesses and consumers billions of dollars annually. Due to privacy concerns, there is a dearth of studies that
analyze actual credit card transactions. In this study, we use machine learning techniques to identify instances
of credit card fraud. In the beginning, regular models are used. After that, we employ a combination of AdaBoost
and majority voting techniques to create a hybrid approach. A publicly accessible data set of credit card
transactions is utilized to assess the performance of the model. Then, a financial institution's actual credit card
data is examined. The resilience of the algorithms is also evaluated by introducing noise into the data samples.
The experimental findings show promise for the majority voting approach as a means of identifying credit card
fraud with high rates of accuracy.