Predicting Users’ Privacy Protection Behavior in Mobile Applications: An Empirical Study Using Machine Learning and Explainable AI
IEEE Access · 2026 · 14 · 77931-77943
DOI: 10.1109/access.2026.3691944 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Access, 2026. Verfasst von Yunfei Wang. Die Arbeit nennt 53 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Scientific Research Project of Anhui Jianzhu University.
Abstract (Originalsprache)
The rapid proliferation of mobile applications has intensified concerns regarding user privacy and personal data protection. Although users can adopt various privacy protection behaviors to mitigate potential risks, the psychological and contextual factors influencing such behaviors remain insufficiently understood. This study presents an empirical, application-oriented prediction study of users’ privacy protection behavior in mobile application environments by integrating Protection Motivation Theory (PMT) with application-level contextual variables. Using cross-sectional survey data collected from 987 respondents, seven machine learning models—including Logistic Regression, Decision Tree, Random Forest, K-Nearest Neighbors, Support Vector Machine, XGBoost, and LightGBM—are employed to predict users’ privacy protection behavior. To enhance model interpretability, SHAP-based explainable artificial intelligence techniques are applied to quantify the relative importance of different predictors. The results show that perceived severity and perceived vulnerability are the most influential factors driving privacy protection behavior, followed by application-level contextual factors such as electronic word-of-mouth, permission request depth, and privacy policy readability. Ensemble learning models demonstrate modestly better predictive performance than traditional models, with XGBoost achieving a slight advantage in test-set AUC, indicating that privacy protection behavior is shaped by complex nonlinear relationships among psychological perceptions and contextual factors. By combining machine learning prediction with explainable artificial intelligence techniques, this study provides empirical and practically interpretable insights into privacy protection behavior and offers implications for privacy-aware mobile application design and digital platform governance.
Förderung
- Scientific Research Project of Anhui Jianzhu University
Zitieren
APA
Wang, Y. (2026). Predicting Users’ Privacy Protection Behavior in Mobile Applications: An Empirical Study Using Machine Learning and Explainable AI. IEEE Access, 14, 77931-77943. https://doi.org/10.1109/access.2026.3691944
Vancouver
Wang Y. Predicting Users’ Privacy Protection Behavior in Mobile Applications: An Empirical Study Using Machine Learning and Explainable AI. IEEE Access. 2026;14:77931-77943. doi:10.1109/access.2026.3691944
BibTeX
@article{Wang2026,
author = {Wang, Yunfei},
title = {Predicting Users’ Privacy Protection Behavior in Mobile Applications: An Empirical Study Using Machine Learning and Explainable AI},
journal = {IEEE Access},
year = {2026},
volume = {14},
pages = {77931--77943},
doi = {10.1109/access.2026.3691944}
}
EUVIMED