End-to-End Customer Churn Prediction Using Machine Learning

---

🚀 Project Overview

This project is a complete end-to-end machine learning pipeline for predicting customer churn. It demonstrates how to transform raw customer data into actionable insights and deploy models that can predict whether a customer is likely to leave a company.

This notebook covers:

• Data Cleaning & Preprocessing
• Exploratory Data Analysis (EDA)
• Feature Engineering
• Training Multiple Machine Learning Models
• Evaluation with Accuracy, F1-Score, and ROC-AUC
• Insights & Recommendations

---

📊 Dataset

• The dataset contains 7043 customers and 21 features, including demographics, subscription details, and payment information.

• Key columns:

  • gender, SeniorCitizen, Partner, Dependents
  • tenure, PhoneService, InternetService
  • OnlineSecurity, DeviceProtection, TechSupport
  • Contract, PaperlessBilling, PaymentMethod
  • MonthlyCharges, TotalCharges
  • Churn (Target column)

• Source: Publicly available customer churn dataset.

---

🧹 Data Preprocessing

• Missing values handled (TotalCharges column).

• Categorical variables encoded using one-hot encoding.

• Features and target separated:

  • Features: X (shape: 7043 x 30)
  • Target: y (shape: 7043)

• Train-test split:

  • Training set: 5634 samples
  • Test set: 1409 samples

---

🤖 Machine Learning Models Trained

The project includes five different models for comparison:

Model	Accuracy	F1-score	ROC-AUC
Logistic Regression	0.8055	0.6029	0.8420
Gradient Boosting	0.7942	0.5672	0.8360
XGBoost	0.7821	0.5621	0.8166
Random Forest	0.7921	0.5594	0.8259
Decision Tree	0.7410	0.5020	0.6610

Best Model: Logistic Regression, based on overall accuracy and ROC-AUC.

---

📈 Visualizations

The notebook contains clear and attractive charts:

• Count plots for categorical features
• Correlation heatmaps
• Distribution plots for numerical features
• Churn rate comparisons across categories

These visualizations help understand patterns in customer churn and provide actionable insights for business strategy.

---

🔧 Requirements

The project uses the following Python libraries:

pandas
numpy
matplotlib
seaborn
plotly
scikit-learn
xgboost
lightgbm
notebook
IPython

---

Dataset Link: https://www.kaggle.com/datasets/mosapabdelghany/telcom-customer-churn-dataset

---

📝 How to Run

1. Clone the repository:

   git clone <your-repo-url>
   

2. Navigate to the project folder:

   cd customer-churn-prediction
   

3. Install dependencies:

   pip install -r requirements.txt
   

4. Open the Jupyter Notebook and run cells sequentially:

   jupyter notebook
   

---

💡 Insights & Recommendations

• Customers on month-to-month contracts are more likely to churn.

• Offering online security and tech support reduces churn.

• High monthly charges increase churn probability.

• Businesses can use this model to target retention strategies for at-risk customers.

---

👤 About the Author

WAQAR ALI — Data Science Student

I am passionate about Machine Learning, Predictive Analytics, and Business-Oriented Data Solutions. This notebook demonstrates a complete end-to-end churn prediction system, designed with professional workflows and interpretable insights.

Connect with me:

📊 Kaggle
💼 LinkedIn
🐙 GitHub
📘 Facebook

---

📌 Notebook created by Waqar Ali — 2025

---