Customer Churn Prediction

This project builds a predictive model to identify customers at risk of leaving a business (churning). The goal is to help stakeholders take proactive retention actions by using machine learning to classify churn based on customer attributes and behavior. It also attempts to analyze the most correlated attributes to identify target areas.

Authors

• Ashir Habib
• Schaff Mudassir Mirza

📁 Project Contents

• CustomerChurnCode.ipynb: Main Jupyter Notebook containing data preprocessing, exploratory data analysis (EDA), model training, evaluation, and interpretation.
• README.md: Overview and instructions for this repository.

📊 Problem Statement

Customer churn is a critical metric for subscription-based or service-oriented businesses. Predicting churn allows companies to reduce revenue loss by targeting at-risk customers with personalized retention strategies.

⚙️ Technologies Used

• Python 3.8+
• Pandas, NumPy – Data manipulation
• Matplotlib, Seaborn – Visualizations
• Scikit-learn – Modeling (Logistic Regression, Random Forest.)
• Jupyter Notebook

🔍 Features

• Data cleaning and preprocessing
• Exploratory data analysis (EDA)
• Feature engineering and selection
• Model building and evaluation
• Classification report and confusion matrix.

📈 Model Performance

• Accuracy: 82%
• Precision: 81%
• Recall: 82%

Project created for the course Machine Learning for Social Scientists at Lahore University of Management Sciences