📊 Telco Customer Churn: A Deep Dive Predictive Analysis

An exhaustive, end-to-end data science project that performs a deep exploratory analysis on a Telco customer dataset to build a high-performance churn prediction model. This repository documents every step, from initial data cleaning to the final model evaluation, which achieved an 0.89 ROC AUC score.

---

📁 Professional Project Structure

Telco-Customer-Churn/
├── data/
│   └── Telco_Customer_Churn.csv  # Raw dataset
├── notebooks/
│   └── analysis_and_modeling.ipynb           # Main Jupyter Notebook
├── visualizations/
│   ├── churn_distribution.jpg
│   ├── demographic_churn_analysis.png
│   ├── contract_churn.png
│   ├── billing_churn.png
│   ├── service_churn_analysis.png
│   ├── tenure_churn.png
│   ├── newplot.png
│   ├── customer_lifetime_value.png
│   ├── roc_auc_curve.png
│   └── confusion_matrix.png
├── requirements.txt                          # Python dependencies
└── README.md                                 # This file

---

🗃️ Dataset Overview

• Source: Kaggle: Telco Customer Churn
• Description: A dataset containing 7,043 customer records with 21 attributes, including demographic data, subscribed services, account information, and the target variable, Churn.

---

📈 All Visualizations & Key Insights

This section provides a complete visual narrative of the exploratory data analysis, with each chart uncovering a different facet of customer churn behavior.

1. Overall Churn Distribution

The analysis begins by establishing the baseline churn rate. The dataset is imbalanced, with 26.5% of customers having churned. This imbalance informs our choice of evaluation metrics like ROC AUC.

2. Demographic Analysis

Insight: While gender is not a factor, Senior Citizens churn at a higher rate. Having a partner and dependents appears to be a stabilizing factor, significantly reducing the likelihood of churn.

3. Contract Term Impact on Churn

Insight: This is the most significant predictor. Customers on Month-to-Month contracts have an extremely high churn rate (over 40%) compared to those on one or two-year contracts, who are far more loyal.

4. Billing and Payment Method Analysis

Insight: Customers using Paperless Billing and paying via Electronic Check are significantly more likely to churn. This may indicate dissatisfaction with the digital billing process or issues related to this specific payment method.

5. Service Subscription Analysis

Insight: Customers without Online Security, Online Backup, and Tech Support services are far more vulnerable to churn. Conversely, having these protective services acts as a strong retention factor. The type of internet service also matters, with Fiber optic customers churning more often than DSL users.

6. Tenure Impact on Churn

Insight: Churn is heavily skewed towards new customers. The first few months are critical, as the churn rate is highest for customers with low tenure. Loyalty builds over time.

7. Correlation Heatmap of Features

Insight: The heatmap confirms the strong relationships identified earlier. Tenure and Contract type have a strong negative correlation with Churn, while Monthly Charges has a positive correlation.

8. Customer Lifetime Value (CLV) Analysis

Insight: An analysis of Customer Lifetime Value reveals that churning customers are often those with lower long-term value. However, the model can be used to prioritize retention efforts on high-CLV customers who are flagged as high-risk, maximizing the financial impact of the retention strategy.

---

🎯 Final Model Performance

The XGBoost Classifier delivered the best performance, validating the insights from our EDA.

• ROC AUC Score: 0.89. This indicates a very high capability to distinguish between churning and non-churning customers.
• Confusion Matrix: The model correctly identifies a significant number of true churners while maintaining high overall accuracy, making it a reliable tool for business action.

---

🛠️ Tech Stack & Setup

• Language: Python 3.x
• Libraries: Pandas, NumPy, Scikit-learn, Matplotlib, Seaborn, XGBoost
• Environment: Jupyter Notebook

How to Run:

1. Clone the repository:

   git clone [https://github.com/YourUsername/Telco-Customer-Churn.git](https://github.com/YourUsername/Telco-Customer-Churn.git)
   cd Telco-Customer-Churn

2. Set up a virtual environment:

   python -m venv venv
   source venv/bin/activate  # On macOS/Linux
   venv\Scripts\activate      # On Windows

3. Install dependencies:

   pip install -r requirements.txt

4. Launch Jupyter and run the notebook:

   jupyter notebook