Classifying Personality Traits from Text Data: A Machine Learning Approach Using Stochastic Gradient Descent for Simplified Jungian Typology-Based Assessment Tool

Currently, In recent years, machine learning approaches have attracted much interest from academic researchers and organizations in various fields, including psychology. Several researchers have developed machine learning models to classify personality based on a person's behavior in online communities over time for several functions such as job search, dating sites, career placement, college major recommendations. However, not all researchers involve psychologists in the model development to ensure the reliability of their models. This can trigger bias due to a lack of understanding in psychology. Generally, people will be asked to answer a personality assessment tool to classify their personality type. Commonly it consists of many questions that may cause a person to lose focus in the answering process. In this research, researchers collaborate with psychologists to develop a machine-learning model based on the simplified personality assessment tool of Jungian typology. Researchers used the Stochastic Gradient Descent algorithm to classify personality based on respondents' answers to the simplified personality assessment tool of Jungian typology. The model evaluation results using the f1-score show that the model developed can classify personality types in the Attitude and Cognitive dimensions with performance reaching 79.05% and 73.99%.