Personality Prediction Using Machine Learning

Abstract

We discover the utility of diverse machine mastering algorithms to are expecting persona sorts primarily based on textual facts using the Myers-Briggs Type Indicator (MBTI) dataset. The dataset incorporates user-generated posts, which are pre-processed through a sequence of herbal language processing (NLP) strategies, including text normalization, stopword elimination, and lemmatization. We appoint the TF-IDF (Term Frequency-Inverse Document Frequency) method to convert the textual information into numerical features. Several classifiers—Gaussian Naive Bayes, Multinomial Naive Bayes, Random Forest, XGBoost, LightGBM, Support Vector Machine (SVM), and Logistic Regression—are skilled and evaluated to predict the MBTI persona kinds. The models are in comparison primarily based on accuracy and certain type reviews. Among the models tested, the XGBoost classifier outperforms others with an accuracy of 67.55%, demonstrating its effectiveness for this multi-elegance text class hassle. This venture highlights the ability of device getting to know in personality prediction from textual records and offers a comparative analysis of diverse type algorithms for this purpose.