📊 Telecom Service Churn Prediction

🔍 A Machine Learning Approach to Predict Customer Churn in the Telecom Sector

---

📌 Project Overview

This project aims to predict customer churn using various machine learning models to help telecom providers proactively retain their customers. Based on real-world data and methodologies from published research, we evaluated model performance and enhanced churn prediction with advanced techniques.

---

🎯 Objectives

• Predict customer churn using models like Decision Trees, Random Forest, SVM, KNN, Naive Bayes, SGD, and ADAM.
• Address class imbalance using SMOTE.
• Perform Survival Analysis to estimate customer lifetime.
• Evaluate models using metrics: Accuracy, Precision, Recall, F1-Score.
• Identify key features influencing churn for better decision-making.

---

📁 Dataset

• Source: Telco Customer Churn Dataset - Kaggle
• Contains customer demographics, account info, and service usage.

---

🛠️ Models Used

• Decision Tree
• Random Forest
• K-Nearest Neighbors (KNN)
• Naive Bayes
• Support Vector Machine (SVM)
• Stochastic Gradient Descent (SGD)
• Multi-layer Perceptron (MLP) with ADAM Optimizer

---

⚖️ Techniques Applied

• Data Preprocessing: Feature engineering, one-hot encoding
• Class Balancing: SMOTE
• Model Optimization: ADAM & SGD
• Evaluation: Confusion Matrix, ROC, Survival Plots

---

🏆 Key Results

• Random Forest achieved ~98% accuracy and strong precision/recall.
• SMOTE significantly improved model fairness across classes.
• Survival analysis (Kaplan-Meier) helped estimate customer retention trends.

---

📆 Timeline

• Week 1–2: Literature Review and Base Paper Analysis
• Week 3–4: Dataset preprocessing & Exploration
• Week 5–6: Model Development & Base Paper Implementation
• Week 7–8: Optimization, Testing, and Final Evaluation

---

📚 References

1. Sharmila K. Wagh et al., Customer churn prediction in telecom sector using machine learning techniques, Results in Control and Optimization, 2024.
2. Abhishek Gaur, Ratnesh Dubey, IEEE Conference 2018, DOI: 978-1-5386-5367-8/18.
3. Telco Customer Churn Dataset - Kaggle
4. Lifelines - Survival Analysis in Python

---

✅ Conclusion

This project showcases the application of machine learning to real-world problems like customer churn in telecom. Using ensemble methods and data balancing techniques, we built a robust and reliable churn prediction model with actionable business insights.