Data Mining Techniques in Handling Personality Analysis for Ideal Customers

Abstract

Background: Personality distinguishes individuals from one another, guides their actions and reactions, and dictates their preferences in many aspects of life, including shopping. Objective: This study determines the characteristics of an ideal customer based on individual personality. Methods: Data mining techniques used in this study are K-nearest neighbour (KNN), linear support vector machine (SVM), and random forest. This study also applies the synthetic minority oversampling technique (SMOTE) to overcome the imbalance in the amount of data. Results: This study shows that the application of the SMOTE and random forest models resulted in 88% accuracy, 79% precision, and 70% recall, which are the highest compared to other models. Conclusion: SMOTE in this research is unsuitable for use in the KNN and linear SVM classification models. Ensemble-based models such as random forest can produce high accuracy when SMOTE is applied for data pre-processing.