A Comprehensive Research of Supervised Learning Strategies Using Machine Learning and Artificial Neural Network Models for Human Personality Trait Classification: A Multi-Model Performance Analysis

Abstract

Computational approaches to personality trait classification are becoming increasingly significant across various disciplines, including psychology, education, and digital behavior analysis. This research investigates the performance of nine supervised learning algorithms, including conventional classifiers e.g., Logistic Regression, Decision Trees, Naive Bayes, K-Nearest Neighbors, and Support Vector Machines, ensemble techniques e.g., Random Forest, Gradient Boosting, and AdaBoost, and Artificial Neural Networks (ANNs) in predicting personality types from structured behavioral data. This dataset consists of features e.g., time spent alone, frequency of social interactions, and digital activity patterns. Preprocessing steps e.g., label encoding are applied to prepare the data for model compatibility. Evaluation is performed using metrics derived from the confusion matrix e.g., accuracy, precision, recall, and F1 score. The findings show that ANNs, Gradient Boosting, Naive Bayes, and SVM consistently achieve the highest performance across all metrics, with an accuracy of 92.93%. Notably, Naive Bayes a relatively simple and computationally efficient model matched the performance of more complex algorithms, suggesting a valuable trade-off between interpretability and predictive power. This highlights the relevance of considering computational efficiency and model transparency when implementing personality classification systems. By offering a comprehensive comparative analysis across different model using structured non-textual behavioral indicators, this research provides new perspectives on designing machine learning-based personality prediction frameworks, particularly in contexts where accuracy and interpretability are equally valued. Furthermore, these findings provide new insights into the potential of lightweight models for scalable psychological profiling, and provides recommendations for future improvements in terms of data diversity, feature expansion, and model explainability.