datathon-v2

Customer Churn Analytics Suite

🎯 Overview

Link to deployed project: https://mihirphalke1-datathon-v2-app-byghe1.streamlit.app/ This project is made as a part of DataThon 2025 held at KJ Somaiya College of Engineering.

The Customer Churn Analytics Model is a comprehensive solution for subscription-based businesses to predict, analyze, and prevent customer churn. Built with Streamlit and powered by advanced machine learning algorithms, this application provides real-time insights and actionable recommendations to improve customer retention.

✨ Features

📊 Customer Clustering

• Advanced K-means clustering with optimized cluster selection
• Interactive 2D and 3D PCA visualizations
• Detailed cluster characteristics analysis
• Feature importance visualization
• Downloadable cluster statistics and analysis

🔮 Churn Prediction

• High-accuracy Random Forest classification model
• Feature importance analysis
• Comprehensive performance metrics
• Interactive confusion matrix visualization
• Exportable prediction results

💡 Retention Strategies

• AI-driven personalized retention recommendations
• Risk-based action prioritization
• Downloadable strategy recommendations
• Customer-specific intervention suggestions

📈 Customer Analysis Dashboard

• Usage pattern analysis
• Customer satisfaction metrics
• Support ticket analysis
• Pricing impact assessment
• Key risk indicators

🛠️ Technical Stack

• Frontend: Streamlit
• Data Processing: Pandas, NumPy
• Machine Learning: Scikit-learn, Lifelines
• Visualization: Plotly
• Statistical Analysis: SciPy

📋 Requirements

streamlit>=1.0.0
pandas>=1.3.0
numpy>=1.21.0
scikit-learn>=1.0.0
plotly>=5.3.0
lifelines>=0.26.0

🚀 Getting Started

Installation

1. Clone the repository:

git clone https://github.com/yourusername/customer-churn-analytics.git
cd customer-churn-analytics

2. Install dependencies:

pip install -r requirements.txt

3. Run the application:

streamlit run app.py

Data Requirements

The application expects a CSV file with the following key columns:

• Customer demographic information
• Usage metrics (ViewingHoursPerWeek, ContentDownloadsPerMonth)
• Payment information (MonthlyCharges, TotalCharges)
• Engagement metrics (UserRating, SupportTicketsPerMonth)
• Churn status

📊 Model Performance

• Clustering Accuracy: Optimized using silhouette scores
• Churn Prediction: >85% accuracy (varies by dataset)
• Feature Importance: Identifies top churn indicators
• Risk Assessment: Real-time probability scoring

🔑 Key Features Explained

Customer Clustering

The clustering module segments customers based on behavior patterns, enabling targeted retention strategies for different customer groups. It uses:

• K-means clustering with automatic optimal cluster selection
• PCA for dimensionality reduction
• Interactive visualizations for cluster analysis

Churn Prediction

Our prediction model uses an ensemble approach to identify at-risk customers:

• Random Forest Classifier with GridSearchCV optimization
• Feature importance analysis
• Robust cross-validation
• Probability-based risk scoring

Retention Strategies

The system provides automated, personalized retention recommendations based on:

• Customer segment analysis
• Historical intervention success rates
• Risk level assessment
• Customer lifetime value