Comparing Data Mining Classification for Online Fraud Victim Profile in Indonesia

Abstract

Classification is one of the most often employed data mining techniques. It focuses on developing a classification model or function, also known as a classifier, and predicting the class of objects whose class label is unknown. Categorizing applications include pattern recognition, medical diagnosis, identifying weaknesses in organizational systems, and classifying changes in the financial markets. The objectives of this study are to develop a profile of a victim of online fraud and to contrast the approaches frequently used in data mining for classification based on Accuracy, Classification Error, Precision, and Recall. The survey was conducted using Google Forms, which is an online platform. Naive Bayes, Decision Tree, and Random Forest algorithms are popular models for classification in data mining. Based on the sociodemographics of Indonesia's online crime victims, these models are used to classify and predict. The result shows that Naïve Bayes and Decision Tree are slightly superior to the Random Forest Model. Naive Bayes and Decision Tree have an accuracy value of 77.3%, while Random Forest values 76.8%.