Customer Churn Analysis – EDA Project

Analyzing Customer Churn Analysis and finding factors that make customers churns and found key areas to avoid churn using Python, NumPy, Pandas, Matplotlib, Seaborn

Overview

This project performs Exploratory Data Analysis (EDA) on a telecommunications company's customer churn dataset. The goal is to uncover patterns and factors that contribute to customer churn and provide actionable business recommendations to reduce churn.

Problem Statements

Dataset

• Source:Dataset-Link
• Rows: 7,043 Columns: 21 features
• Target Variable: Churn (Yes = customer left, No = customer stayed)

Tools and Technologies

• Find insights using Python (NumPy, Pandas, Matplotlib, Seaborn)

Steps Performed

• Data Cleaning

• Converted TotalCharges to numeric
• Removed 11 missing rows (0.15% of data)
• Created tenure groups for better segmentation
• Dropped irrelevant columns (customerID, raw tenure)

• Exploratory Data Analysis

• Demographic analysis (gender, senior citizen, dependents)
• Service usage patterns (internet type, online security, tech support)
• Contract & billing analysis
• Financial analysis (monthly charges, total charges)
• Tenure vs churn patterns
• Correlation analysis

• Visualization Tools Used

• Matplotlib & Seaborn for bar charts, KDE plots, and correlation heatmaps

Keyinsights

1. Senior citizens are more likely to churn.
2. People with no partner is high churn rate.
3. People with no dependency has more churn rate.
4. People using internet services with Fibre Optics has high churn rate.
5. People with no online security has high churn rate.
6. People with no online backup has high churn rate.
7. People with no device protection has high churn rate.
8. People with no tech support has high churn rate.
9. People with Month-to-Month Contract has highest churn rate.
10. People with high paperless billing has high churn rate.
11. People with Electronic Check payment method has high churn rate. (i.e. 45%)
12. People with tenure group between 1-12 month has highest churn rate.
13. Monthly Charges and Total Charges are positively correlated
14. Churn is high when Monthly Charges are high and Churn is low when Monthly Charges are low
15. People with No Online Security is high Churners

Result and Cunclusions

1. Promote long-term contracts with discounts.
2. Bundle tech support and online security with internet services.
3. Focus retention efforts on first-year customers.
4. Offer special pricing for high monthly charge customers to reduce early churn.
5. Encourage automatic payment options to reduce churn.

Future Work

• Interactive Dashboard Creation
• Using Power BI to:
• Monitor churn rate over time
• Drill down into contract type, tenure, and services
• Display KPIs like Churn %, Revenue Lost, Top Risk Segments

Contact : LinkedIn E-mail: [email protected]