Machine learning-based predictive models for digital behavioral analysis

Introduction: The rise of digital technology use in Ecuador has produced large volumes of data on user behavior. In this context, machine learning models provide an effective way to analyze and predict digital behavior patterns, supporting informed decision-making in fields such as marketing, education, and public policy. Methods: A quantitative, non-experimental, cross-sectional methodology was used. A Random Forest model was applied to a simulated dataset based on parameters from the National Institute of Statistics and Censuses (INEC). The analysis focused on variables such as age, internet connection frequency, device type, and type of content consumed. Data were processed using Python and specialized machine learning libraries. Results: The model achieved 91,3 % accuracy in classifying digital user profiles. The most predictive variables were weekly connection frequency, type of digital content, and age. Distinct behavioral patterns were identified among age groups, allowing for relevant personalized strategies. Conclusions: The results demonstrated the effectiveness of machine learning in classifying and understanding digital behavior in Ecuador. This approach proves useful for designing more effective and ethically responsible digital interventions, as long as data privacy and protection principles are upheld.