Predicting Smartphone Addiction Through Machine Learning

                                                          

                                     

                                       ABSTRACT

        Smartphone addiction is a growing issue, impacting mental health and productivity. This project leverages machine learning to predict addiction risk using behavioral data from our own dataset and user surveys. Our model’s accuracy will be evaluated using precision, recall, and F1 score to ensure reliable classification.  Beyond prediction, we will assess user engagement by tracking interactions and analyzing behavioral changes through pre- and post-surveys. This will help determine the system’s effectiveness in raising awareness and encouraging healthier smartphone use.  User feedback will be key in refining our platform, ensuring personalized recommendations for better digital well-being. By combining  insights with real-world impact, our goal is to create a system that not only detects addiction risk but also promotes responsible smartphone usage.

                                                         ARCHITECTURE DIAGRAM

                                       

PROPOSED SYSTEM

• User-Centered Analysis: Focuses on real user data and survey responses  
• Behavior-Based Categorization: Assesses screen time, app usage, and survey responses to classify users into Low, Moderate, or High-risk levels.  
• Insight-Driven Approach: Helps users reflect on their habits instead of relying on automated decision-making.  
• Simple & Accessible Interface: Built with Flask, HTML, CSS, and JavaScript for ease of use.  
• Personalized Guidance: Provides practical strategies and recommendations for healthier smartphone use without imposing solutions.

                                                             RESULTS

                                        

   

    

     

                                                

                                                   

We use machine learning as a supportive tool to analyze smartphone usage patterns. We tested five models—Random Forest, XGBoost, KNN, SVM, and Logistic Regression—while ensuring fair representation of all risk levels using SMOTE (Synthetic Minority Over-sampling Technique). 

Among these, XGBoost and Random Forest provided the most reliable insights, achieving 94% accuracy in categorizing users into Low, Moderate, or High-risk groups. However, our system focuses on user awareness. 

Instead of just automated classification, we provide practical tips and recommendations based on survey responses, helping users reflect on their habits and take control of their smartphone usage in a way that suits their lifestyle.