Digital Behavior and Mental Health: A Study on Concentration Classification Using Machine Learning

This project is a machine learning project for beginners. It uses logistic regression to classify and predict mental health concentration levels by analyzing digital behavior data such as screen time, notification frequency, and app switching. The goal is to help users understand the impact of digital habits on concentration and provide data support for personalized health management.

In the digital age, smart devices are deeply integrated into daily life, but excessive use may lead to problems such as distraction and anxiety. This project aims to establish a correlation model between digital behavior patterns and concentration levels using machine learning technology, answer which digital behaviors affect concentration and to what extent, and provide users with data-driven insights.

Data Collection and Feature Engineering

Focus on three types of indicators: screen time (duration of each app), notification frequency (number of various notifications), and app switching frequency (number of switches between apps).

Model Selection

Logistic regression algorithm is chosen because of its strong interpretability, high computational efficiency, friendliness to beginners, and stable performance on small datasets.

Data Processing Flow

Use Python libraries (Pandas, NumPy, Matplotlib, Scikit-learn) for data loading, cleaning, normalization, model training, and evaluation.

• Beginner-friendly: Detailed documentation lowers the entry barrier, allowing new users to run the application according to guidelines.
• Complete workflow: Covers the entire lifecycle of data collection, exploration, feature engineering, model training, evaluation, and result interpretation.
• Open-source collaboration: Open-sourced under the MIT license; community feedback and improvements are welcome.

• Personal digital health management: Monitor changes in digital behavior and adjust habits (e.g., notification settings).
• Digital detox evaluation: Quantify the impact of behavioral changes on concentration.
• Academic support: Provide a reference framework for digital health and behavioral psychology research.

Limitations

• Data privacy: Need to balance analysis needs and privacy protection.
• Causal relationship: The model reveals statistical correlations rather than causality.
• Individual differences: A unified model is difficult to capture the sensitivity of all users.

Improvement Directions

• Introduce more features such as sleep and exercise.
• Try complex models like random forests to improve accuracy.
• Develop personalized models and intervention recommendation functions.

Provides beginners with:

• Classification problem practice: Master the supervised learning classification process.
• Feature engineering experience: Extract effective features.
• Model evaluation methods: Understand metrics like accuracy.
• Data visualization: Create charts using Matplotlib.
• Python ecosystem: Familiarize with tools like Pandas and Scikit-learn.

Although this project is small in scale, it touches on mental health issues in the digital age and demonstrates how technology can help understand oneself and improve life. In today's attention economy-dominated world, managing digital behavior is an important skill, and such tools provide an objective perspective to help balance digital convenience with physical and mental health.