Customer Churn Prediction

Project Overview

This project aims to predict customer churn using machine learning techniques. By leveraging data analytics, we derive actionable insights to enhance customer retention strategies and improve business decision-making.

Team Members

• Akshada Girish Malpure (agmalpur)
• Rishi Dange (radange)
• Rucha Mahesh Kulkarni (rkulkar5)

Course

CSC522 Automated Learning and Data Analysis

Dataset

We used the Telco Customer Churn dataset from Kaggle:
Telco Customer Churn Dataset

• 7043 customer records
• 21 features
• 70-30% train-test split

Key Features

• Personal Information: Gender, age, dependents
• Service Details: Phone, internet, security features
• Billing Information: Monthly charges, payment methods, tenure
• Target Variable: Churn (Binary - customer left or not)

Methodology

1. Data Preprocessing

• Handling Missing Values: Mean/mode imputation
• Feature Selection: Dropped non-informative columns (e.g., customerID)
• Encoding & Scaling:
  • Label encoding for categorical variables
  • Standard scaling for numeric values
• Class Imbalance Handling: Used SMOTE to balance churn vs. non-churn classes

2. Feature Engineering

• Identified key features using heatmaps and domain knowledge
• Created tenure-based customer groups
• Engineered features to improve model interpretability

3. Model Training & Evaluation

Trained multiple machine learning models:

• Logistic Regression
• Random Forest
• Gradient Boosting
• XGBoost
• LightGBM
• Decision Tree
• KNN
• Naive Bayes
• SVM

Evaluation Metrics:

• Accuracy
• Precision, Recall, F1-score
• AUC-ROC

Best Performing Models

• Random Forest, LightGBM, Gradient Boosting achieved around 85% accuracy after hyperparameter tuning.

Key Insights

• Contract type, tenure, and payment methods are the most significant factors affecting churn.
• Short-tenured customers are at the highest risk of churning.
• Customers using automated payment methods are less likely to churn.

Future Scope

• Real-Time Prediction: Deploy a model for real-time customer churn analysis.
• Automated Machine Learning: Optimize models using AutoML techniques.
• Integration with Behavioral Data: Enhance predictions by including customer interaction data.

Conclusion

Machine learning enables businesses to predict customer churn with high accuracy. By implementing targeted retention strategies, companies can reduce attrition and improve customer loyalty.