Feature Selection Analysis for Diagnosing Narcissistic Personality Disorder (NPD) Using Principal Component Analysis and the Naïve Bayes Model

Abstract

The mental health illness known as narcissistic personality disorder (NPD) affects a person's capacity to preserve harmonious social interactions. Early diagnosis plays a crucial role in providing timely intervention and treatment. This study examines the effectiveness of Principal Component Analysis (PCA) for feature selection in diagnosing NPD using the Naïve Bayes algorithm. The dataset utilized in this research was sourced from Open Psychometrics via Kaggle, followed by preprocessing, including data cleaning and dimensionality reduction through PCA. This study compares the performance of three Naïve Bayes models, Gaussian, Bernoulli, and Multinomial, to identify the most suitable classification approach. The findings reveal that Gaussian Naïve Bayes, when integrated with PCA, achieves the highest accuracy (91%), surpassing Bernoulli Naïve Bayes (80%) and Multinomial Naïve Bayes (69%). Implementing PCA significantly enhances computational efficiency and improves classification performance by eliminating irrelevant features and reducing data dimensionality. These results suggest combining PCA with Gaussian Naïve Bayes is a promising strategy for automating NPD diagnosis. Additionally, this study highlights the potential of machine learning in mental health evaluation and establishes the framework for further studies on hybrid models or other methods to improve prediction accuracy.