Mental Health Risk Prediction App

• This Streamlit-powered web app helps predicting the risk of mental health issues based on social media usage and demographic information using multiple machine learning models.
• From data preprocessing to model deployment, this project showcases a full data science pipeline and an interactive web app

Key Highlights

• Complete ML Workflow: Data cleaning, EDA, feature engineering, model training, and evaluation.
• Model Portfolio: Logistic Regression, Decision Tree, Random Forest, Naive Bayes, and XGBoost.
• Robust Evaluation: Accuracy, Precision, Recall, F1-Score, ROC-AUC, and Confusion Matrix.
• Interactive Deployment: Streamlit app with login authentication for real-time predictions.
• Social Impact: Early detection of mental health risks through data-driven insights.

Installation & Usage

1. Clone the repo

   git clone https://github.com/yourusername/mental-health-predictor.git
   cd mental-health-predictor
   

2. Install dependencies

   pip install -r requirements.txt
   

3. Explore the analysis notebook

   jupyter notebook mental_health_analysis.ipynb
   

4. Launch the Streamlit app

   streamlit run Home.py
   

How It Works

1. Data Preprocessing: Handle missing values, encode categorical features, normalize inputs.
2. Exploratory Data Analysis: Visualize data distributions, correlations, and platform usage insights.
3. Model Training & Comparison: Train five algorithms and compare performance metrics.
4. Evaluation: Generate and display detailed metrics and confusion matrices.
5. Deployment: Interactive web app lets users input their info and receive a risk prediction.

Summary

• Built and deployed an ML pipeline to predict mental health risks from social media usage and demographic data.

• Achieved 71.7% accuracy with tuned logistic regression, outperforming more complex models like tree-based models.

• Delivered insights on model interpretability and dataset limitations, showing the values of simpler models in certain domains