Identifying 16 Personalities Based on Daily Writing by Random Forest and Logistic Regression

Abstract

The Myers-Briggs Type Indicator (MBTI) is a self-reported personality test based on psychological theory proposed by Carl Gustav Jung, elaborated and developed by Katharine Cook Briggs and Isabel Briggs Myers. It separates human’s personality into 4 dimensions: mind, energy, nature and tactics, 2 directions are provided in each dimension, and it results 16 distinct personalities. Through the exquisite figures design of each personality and extensive advertisement, the test has immediately attracted a huge amount of participants among the youth. Finally, the MBTI test gradually creates a new trend of labeling and judging people according to their personality. Current studies have predicted the personality through linear SVM, Neural network, and K-means. This study devotes to utilize Logistic Regression model and Random forest model to predict based on the daily conversation. As a result, the accuracy that the author gets is relatively high comparing with similar works, and the best accuracy of 4 dimensions (Mind, Energy, Nature, Tactics) in author’s model reaches 78.67%, 85.82%, 58.73% and 63.0%.. The purpose of this study is examining the practicability of implementing machine learning in the psychological field, meanwhile, extends the methods of predict one’s personality. Studies in this field can be further used to help analyze a people, and assist the doctors to treat psychological problems.