ICIC Express Letters ICIC International c ⃝2022 ISSN 1881-803X Volume 16, Number 2, February 2022 pp. 117–125 MACHINE LEARNING ALGORITHMS EXPLORATION FOR PREDICTING PERSONALITY FROM TEXT Andry Chowanda 1 , Derwin Suhartono 1 , Esther Widhi Andangsari 2 and Kamal Zuhairi bin Zamli 3 1 Computer Science Department, School of Computer Science 2 Psychology Department, Faculty of Humanities Bina Nusantara University Jl. K. H. Syahdan No. 9, Kemanggisan, Palmerah, Jakarta 11480, Indonesia { achowanda; dsuhartono; esther }@binus.edu 3 Faculty of Computing University Malaysia Pahang Gambang, Kuantan 26300, Malaysia kamalz@ump.edu.my Received March 2021; accepted June 2021 Abstract. This research explores several best machine learning algorithms to build a model for personality prediction from the text. Moreover, extensive feature sets were also explored to determine the best features to represent the dataset. The personality model implemented in this research was the Myers-Briggs Type Indicator (MBTI) model, where there is no much research done to automatically predict the MBTI personality type using machine learning. The dataset used in this research was from the (MBTI) Myers-Briggs Personality Type Dataset. Oversampling and undersampling techniques were applied to the dataset to make the dataset more balanced. The Artificial Neural Networks algorithm achieved the best result with the score of 76.3% and 77.5% for accuracy and F1 score, respectively. Keywords: Personality prediction, Machine learning, Features exploration, MBTI 1. Introduction. Personality characterizes the individuals’ characteristics, patterns of thought, emotions and behaviour. Personality is unique to each individual and can be recognized by using specific tests conducted by experts. Recently, machine learning is im- plemented to build an automatic personality prediction from people’s face, handwriting, text, and voice prosody. The prediction model can be used in many cases such as job appli- cation, marketing segmentation, and enhancing user experiences by displaying the user’s preferences when interacting with the application or system. Several personality models and tools exist, such as the Big Five Personality, DISC Profile test, and the MBTI (Myers- Briggs Type Indicator). There is not much research in predicting personality from text based on the MBTI model by using machine learning. Hence, this research aims to explore several machine learning techniques to build a model that can predict the MBTI model of personality by using text modality. MBTI is a personality psychological instrument con- structed by Myers and Briggs based on Jung theory [1]. It focuses on the character and dynamic of each personality type explained in it. This instrument does not use alone as a personality profiling assessment but also as a complementary or supporting instrument to explain other aspects, such as career profiling [2] and students’ learning style [3]. Users need to understand the four dichotomies are constructed in MBTI. Everyone has these four dichotomies, as Jung explained in his theory [1]. The perception of activities is rep- resented by Sensing (S) and Intuition (I); rational functions are represented by Thinking DOI: 10.24507/icicel.16.02.117 117