📞 Telecom Churn Prediction

Predicting customer churn is critical for telecom companies aiming to retain valuable customers and reduce revenue loss. This project builds a machine learning pipeline to analyze telecom customer data and predict the likelihood of churn, using a Logistic Regression model and advanced preprocessing techniques.

---

🚀 Project Overview

• Goal: Predict which customers are likely to churn based on their usage and account features.
• Techniques:
  • Data preprocessing and feature selection
  • Handling class imbalance with SMOTE
  • Logistic Regression modeling
  • Feature importance analysis
  • Model evaluation (accuracy, confusion matrix, classification report)

---

📂 Dataset

Source:

• Rows: 3,333
• Features: 11 (including Churn, AccountWeeks, ContractRenewal, DataPlan, DataUsage, CustServCalls, DayMins, DayCalls, MonthlyCharge, OverageFee, RoamMins)

---

Project Notebook

telecom_churn.csv

🛠️ Project Structure

---

📊 Key Steps

1. Data Preprocessing

   • Handle missing values
   • Encode categorical variables
   • Scale features

2. Class Imbalance Handling

   • Stratified train-test split
   • Synthetic Minority Oversampling Technique (SMOTE)

3. Modeling

   • Logistic Regression (with class_weight and/or SMOTE)
   • Feature importance analysis

4. Evaluation

   • Accuracy, Confusion Matrix, Classification Report

---

🖼️ Example Visualizations

• Feature Importance using Linear Regression Model
  • Feature Importance using XGBOOST Model

⚙️ Requirements

• Python 3.8+
• pandas
• numpy
• scikit-learn
• imbalanced-learn
• matplotlib
• seaborn

---

🤝 Contributing

Contributions are welcome! Please fork the repository and submit a pull request.

---

📄 License

This project is licensed under the MIT License.

---

📬 Contact

For questions or feedback, please open an issue or contact [[email protected]].

---

Empowering telecoms with data-driven retention strategies!

---

🏁 Getting Started

1. Clone the repository