📈 Telecom Customer Churn Prediction

Machine learning project for predicting and understanding customer churn in the telecom industry.

---

📌 Introduction

This project analyzes telecom customer data to identify factors that influence churn and builds machine-learning models to predict customers likely to leave. The goal is to help businesses improve retention and reduce revenue loss.

---

📂 Dataset Overview

The dataset contains 7,043 customer records with:

• Demographics: gender, senior citizen, dependents
• Account details: tenure, contract, payment method, billing
• Services: phone, internet, backup, security, streaming, tech support
• Target variable: Churn (Yes/No)

Example columns:

customerID | gender | tenure | InternetService | Contract | MonthlyCharges | Churn

---

🧹 Data Cleaning

• Dropped irrelevant column: customerID
• Converted TotalCharges to numeric
• Handled missing values and tenure=0 cases
• Standardized categorical values (e.g., SeniorCitizen)

---

📊 Exploratory Data Analysis (EDA)

Key EDA insights (visuals recommended):

• Churn distribution
• Tenure vs churn patterns
• Contract type impact
• Monthly charges comparisons
• Service usage behavior

(Insert visualizations or screenshots here.)

---

🧽 Data Preprocessing

• Label encoding for categorical variables
• Standardizing numerical features: tenure, MonthlyCharges, TotalCharges
• Train–test split
• Scaling with StandardScaler
• Key practices: split before preprocessing/oversampling, fit transformers on training only, oversample training set only.

---

🤖 Modeling

Models used:

• Logistic Regression
• KNN
• SVC
• Decision Tree
• Random Forest
• Gradient Boosting
• AdaBoost
• XGBoost
• CatBoost
• Voting Classifier (ensemble)

Evaluation metrics: Accuracy, Precision, Recall, F1-score, ROC-AUC, Confusion Matrix.

(Insert model comparison tables and confusion matrices.)

---

🏆 Results

• Identified key churn drivers

• Best-performing model: (Insert your final model and score here)

  • Example: Gradient Boosting — Accuracy: 80%, ROC-AUC: 0.82

---

🛠 Technologies Used

• Python
• Pandas, NumPy
• Matplotlib, Seaborn, Plotly
• Scikit-learn
• XGBoost, CatBoost

---

▶️ How to Run

git clone https://github.com/wonderakwei/Telecom-Customer-Churn-Analysis.git
cd Telecom-Customer-Churn-Analysis
pip install -r requirements.txt
jupyter notebook Telecom_Churn_Prediction.ipynb

---

🚀 Future Improvements

• Streamlit web app deployment
• SHAP feature importance
• More hyperparameter tuning
• Interactive churn dashboard

---

👤 Author

Wonder Akwei Data Analyst | Machine Learning Enthusiast | Fintech Operations

Email: akweiwonder@outlook.com

LinkedIn: https://www.linkedin.com/in/wonderakwei/

---