Özet:

In the study, it is aimed to contribute to the fraud detection process with an application in which machine learning method is used to reduce the damage caused by fraudulent disbursements, which is included in the fraud tree of the Association of Certified Fraud Examiners (ACFE). To achieve the desired outcomes, a Python application system is developed for this purpose. In the study, an artificial data set containing normal transactions and fraudulent transactions of a bank, was used. Using the Decision Tree technique, which was selected as the chosen method, the main model was developed using a data set with known class labels, and then the model was evaluated using unlabeled data. The model of the decision tree technique achieved 97,1% accuracy, 98,4% f1-score, 98,9% precision and 98% sensitivity. The study is open to improvement in terms of reducing the erroneous class labels produced by the decision tree technique during the estimation phase and can be improved by comparing it with other techniques.

Anahtar Kelimeler: