<![CDATA[Handwriting represents a non-verbal behavioral cue that can be used to identify individual personality characteristics. However, conventional manual assessment of personality scores is inherently subjective and requires specialized expertise, highlighting the need for a more objective, efficient, and consistent automated approach. This study aims to develop a system for predicting personality trait scores from handwriting images. The personality assessment is based on the Big Five Personality model, comprising Openness, Conscientiousness, Extraversion, Agreeableness, and Neuroticism. The research methodology adopts an adapted Cross-Industry Standard Process for Data Mining (CRISP-DM) framework, including the stages of business understanding, data understanding, data preparation, model development, evaluation, and system implementation. The proposed system employs the VGG19 deep learning architecture and compares the performance of three optimization algorithms: Adam, Stochastic Gradient Descent (SGD), and RMSProp. The evaluation results demonstrate that RMSProp achieved the best performance on the validation dataset, with a Mean Squared Error (MSE) of 0.0128, Mean Absolute Error (MAE) of 0.0930, Root Mean Squared Error (RMSE) of 0.1133, Pearson Correlation Coefficient (PCC) of 0.4648, and an accuracy of 90.70%. The VGG19 model optimized with RMSProp was subsequently deployed in a mobile application capable of receiving handwriting image inputs and generating predicted Big Five personality trait scores. These findings demonstrate that integrating deep learning with mobile applications offers a practical and effective solution for predicting Big Five personality trait scores from handwriting images.]]>
Copyrights © 2026
