A Comparative Study of Machine Learning Algorithms for Classifying Personality Type based on Carl Jung's Cognitive Function

According to Carl Jung, every human being has a personality preference from each dichotomy dimension of cognitive function to live life. Cognitive function is mental attitudes and actions that always coexist with humans. This theory is the basis for developing personality assessment instruments, such as the Myers-Briggs Type Indicator (MBTI). On the other hand, Machine Learning (ML) approach is often used in psychology to classify personality types automatically. However, selecting the most suitable algorithm and developing the instruments for data collection is not an easy task. This paper aims to conduct experiments choose the most appropriate algorithm for classifying personality types based on Carl Jung's theory of cognitive function. In collaboration with a psychologist, we adapt Carl Jung's theory of cognitive function into a text-based personality assessment instrument consisting of several items. The experimental results show that the Stochastic Gradient Descent algorithm with hyperparameter tuning consistently yields good performance in carrying out cognitive function personality type classification tasks.