Telecom Customer Churn Analysis

End-to-End Data Analysis & Machine Learning Project

Problem Statement

A telecom company is losing customers every month. This project identifies why customers churn, who is most at risk, and builds a machine learning model to predict churn — enabling the business to take proactive action before losing customers.

---

Project Goals

1. Perform in-depth EDA to uncover churn patterns
2. Identify the top factors driving customer churn
3. Build and compare 4 ML models to predict churn
4. Visualize findings in a Power BI Dashboard
5. Provide actionable business recommendations

---

Project Structure

telecom-churn-analysis/
│
├── Telecom_churn_analysis.ipynb         # Complete analysis + ML models
│
├── telecom_customer_churn.csv           # Main dataset (7,043 customers × 38 features)
├── telecom_data_dictionary.csv          # Column descriptions
├── telecom_zipcode_population.csv       # ZIP code population data
│
├── README.md
└── .gitignore

---

Dataset

Property	Value
Rows	7,043 customers
Columns	38 features
Target Variable	Customer Status (Churned / Stayed / Joined)
Churn Rate	~26.5%

---

Analysis Performed

Exploratory Data Analysis (EDA)

• ✅ Overall churn rate & distribution
• ✅ Churn reasons & categories (why customers leave)
• ✅ Contract type vs churn
• ✅ Tenure vs churn (grouped by tenure range)
• ✅ Monthly & total charges vs churn
• ✅ Gender, age, partner & dependents analysis
• ✅ Internet service & type analysis
• ✅ Feature correlation heatmap
• ✅ ZIP code population analysis

Machine Learning (4 Models Compared)

Model	Notes
Logistic Regression	Baseline model
Decision Tree	Interpretable rules
Random Forest	Ensemble, high accuracy
Gradient Boosting	Best performance

• ✅ Feature engineering & label encoding
• ✅ StandardScaler normalization
• ✅ Cross-validation (5-fold)
• ✅ ROC-AUC comparison
• ✅ Confusion matrix
• ✅ Feature importance chart

---

Key Findings

#	Insight
1	~26% of customers churned — major revenue impact
2	Competitor offers are the #1 churn reason
3	Month-to-month customers churn 3× more than yearly contract customers
4	New customers (0–12 months) are at the highest risk
5	Higher monthly charges are associated with higher churn
6	Customers without dependents or partner churn more

---

Business Recommendations

1. Incentivize long-term contracts — offer discounts to switch from month-to-month plans
2. Early retention program — proactive support for customers in their first 12 months
3. Competitive pricing — review pricing strategy since competitors are the #1 churn driver
4. Deploy ML model — flag at-risk customers monthly for targeted outreach
5. Loyalty rewards — recognize and reward long-tenure customers

---

Technologies Used

• Python 3.10 — core language
• Pandas & NumPy — data manipulation
• Matplotlib & Seaborn — visualization
• Scikit-learn — ML models, evaluation metrics
• Power BI — interactive dashboard

---

How to Run

1. Clone this repository:

   git clone https://github.com/your-username/telecom-churn-analysis.git
   cd telecom-churn-analysis

2. Install dependencies:

   pip install pandas numpy matplotlib seaborn scikit-learn jupyter

3. Launch Jupyter Notebook:

   jupyter notebook

4. Open Telecom_Churn_Analysis.ipynb and run all cells.

---