Mental Health & Social Media Balance Dataset | Python, Pandas, EDA
Performed user behavior analysis EDA to identify mental well-being patterns, high-risk user segments, and behavioral thresholds, and translated them into actionable recommendations for a digital wellness product.
MindEase is developing a Digital Balance Coach to provide personalized guidance on screen-time management, sleep hygiene, and digital detox habits.
The product team lacks clarity on:
- Which user groups are at the highest mental health risk
- Which behaviors most strongly influence stress, sleep, and well-being
- What thresholds should trigger personalized recommendations
This analysis provides the behavioral and segment-level insights required for feature design and prioritization.
As the data analyst, the goals are to:
- Which behaviors (screen-time, sleep quality, addiction score) are most correlated with poor mental well-being?
- Which user segments (age, gender, usage levels) show the highest risk of stress or low mood?
- At what screen-time threshold do we see significant declines in mental well-being?
- What patterns predict poor sleep or high stress?
- What product recommendations can reduce mental health risks for each segment?
- Screen-time is the strongest driver of stress. Levels rise sharply after 6.5 hours/day.
- Sleep quality is tightly linked to stress. Poor sleep (<5) consistently predicts high stress.
- Exercise improves well-being. Users exercising ≥3 times/week show higher happiness and lower stress.
- Users with 0 disconnect days show higher stress and lower sleep.
- High-risk segments:
- Age 16–24: highest screen-time and stress
- Users with >7 hrs/day screen-time
- Low-activity users (<2 workouts/week)
- Screen-time coaching: Trigger interventions when usage exceeds 6.5 hrs/day.
- Sleep routines: Provide wind-down tips and reduce night notifications for poor sleepers.
- Digital detox nudges: Encourage breaks for users with 0 disconnect days.
- Promote micro-activity nudges for low-exercise users.
- Activity boosters: Micro-activity reminders for users exercising <2x/week.
- Age-based personalization:
- 16–24: strict screen-time boundaries
- 25–34: stress management
- 35+: sleep optimization
This section outlines the tools, techniques, and step-by-step analysis workflow used to generate the project insights.
Source: Kaggle - Mental Health & Social Media Balance Dataset
Scope : 500 users
Key columns include:
- Daily Screen Time (hrs)
- Sleep Quality (1–10)
- Stress Level (1–10)
- Exercise Frequency (per week)
- Days Without Social Media
- Happiness Index (1–10)
- Age, Gender, Platform
- Python 3.10
- Pandas, NumPy
- Matplotlib, Seaborn
- Correlation analysis
- Segmentation analysis
- Clone the Repository
git clone https://github.com/zaraaxdata/digital-wellness-app-analysis.git
cd digital-wellness-app-analysis
- Create Environment
conda create -n eda_env python=3.10
conda activate eda_env
pip install -r requirements.txt
- Launch Notebook
jupyter notebook
- Data Inspection & Cleaning
Verified data types, missing values, duplicates. - Univariate Analysis
Explored distributions of screen-time, stress, sleep, and exercise patterns. - Bivariate Analysis
- Correlation heatmap
- Screen-time vs stress
- Sleep vs stress
- Exercise vs happiness
- Days without social media vs well-being etc..
- Segmentation Analysis
- Age groups (16–24, 25–34, 35–44, 45+)
- Gender-based behavior diffe
- Screen-time risk tiers (low/medium/high)
- Answers to Business Questions Converted patterns into actionable product recommendations.
- Build Streamlit dashboard for product stakeholders
- Build Stress prediction model
- Build Recommendation engine
If you find this project helpful, don’t forget to give a ⭐.