Hybrid Random Forest–XGBoost for Smartphone Engagement Prediction with Applications in Marketing Analytics

This study presents a machine learning framework for predicting smartphone engagement intensity using behavioral smartphone usage data. The study utilized the Smartphone Usage and Addiction Analysis dataset containing 7,500 user records with demographic, behavioral, lifestyle, and engagement-related variables. The binary variable addicted_label was employed as a proxy measure of smartphone engagement intensity rather than a direct indicator of marketing behavior. Seven machine learning models were developed and evaluated, including Logistic Regression, Decision Tree, Support Vector Machine, Naïve Bayes, Random Forest, XGBoost, and a proposed Hybrid Random Forest–XGBoost model. Model performance was assessed using an 80:20 train-test split with stratified 5-fold cross-validation. Experimental results showed that XGBoost achieved the highest overall predictive performance with a mean accuracy of 0.9433, an F1-score of 0.9591, and an ROC-AUC of 0.9906. Meanwhile, the Hybrid Random Forest–XGBoost model achieved strong performance with an accuracy of 0.9427, an F1-score of 0.9581, an ROC-AUC of 0.9896, and the highest precision of 0.9929. Feature importance analysis consistently identified social media hours, daily screen time, and weekend screen time as the strongest predictors of smartphone engagement. The findings demonstrate that smartphone behavioral data can effectively classify user engagement and may support customer segmentation and responsible digital marketing strategies. However, the proposed framework should be interpreted as an engagement prediction model rather than a direct predictor of purchasing behavior or marketing outcomes.

@article{Liwanag2026,
  title     = {Hybrid Random Forest–XGBoost for Smartphone Engagement Prediction with Applications in Marketing Analytics},
  journal   = {International Journal of Advanced Computer Science and Applications},
  volume    = {17},
  number    = {7},
  year      = {2026},
  publisher = {The Science and Information Organization},
  author    = {Ace John Mark P. Liwanag},
  doi       = {10.14569/IJACSA.2026.0170743},
  url       = {https://doi.org/10.14569/IJACSA.2026.0170743}
}