Fraud Detection in Sales of Distribution Companies Using Machine Learning

Abstract

The sales department of a distribution company is one of the places where fraud often occurs. This fraud occurs in various ways and causes massive losses for the company. These frauds have certain patterns. The patterns that occur in these practices are studied by the company's internal auditor experts.  The experience of these experts is processed into a system called the Expert System. The support of a technology-based tool is needed in order to detect sales fraud early. The purpose of this research is to be able to provide benefits for companies with early detection of fraud in the sales department. At the time this research was conducted, researchers had not found similar research with the same object.
In this research, a comparison of various machine learning algorithm models will be carried out with the aim of knowing whether using machine learning technology can help detect fraud with a high accuracy value. The algorithm method used is supervised learning method. The algorithm models to be compared are Decision Tree, K-Nearest Neighbor, Random Forest, SVM and Logistic Regression. It is expected that by using machine learning technology, fraud can be detected early, so that the level of loss and risk of sales can be minimized.