Project 2 — Customer Churn Analysis¶
Predict which customers are about to leave and understand why — before it's too late.
Overview¶
Customer churn is one of the most expensive problems in any subscription or repeat-purchase business. Acquiring a new customer costs 5-7× more than retaining an existing one. This project builds a complete churn analysis: from raw data to a prioritised list of at-risk customers and actionable retention recommendations.
Business Objective¶
Reduce monthly churn rate from 8.2% to under 5% by identifying at-risk customers at least 30 days before they churn, enabling targeted retention interventions.
Tools and Skills¶
| Tool | Used for |
|---|---|
| Python (Pandas) | Data loading, cleaning, feature engineering |
| SQL | Cohort analysis, churn rate queries, segmentation |
| matplotlib / seaborn | Churn distribution, cohort heatmaps, funnel charts |
| scikit-learn | Logistic regression churn model (conceptual) |
Difficulty¶
Intermediate — requires comfortable Pandas skills and basic SQL. The ML section is conceptual (understanding, not implementation from scratch).
Dataset¶
- Source: Telecom churn dataset (IBM/Kaggle) — 7,043 customers, 21 features
- Download: Kaggle Telco Customer Churn
- Schema details: dataset-guide
Project Files¶
| File | Description |
|---|---|
| business-problem | Business context and questions to answer |
| dataset-guide | Dataset schema, column descriptions, data quality notes |
| data-cleaning | Cleaning steps with Python code |
| sql-analysis | SQL queries answering the business questions |
| dashboard-design | What to build in Power BI / Tableau |
| insights-and-recommendations | Findings and actionable recommendations |
| interview-questions | Questions an interviewer might ask about this project |
Learning Outcomes¶
After completing this project you will be able to: - Calculate churn rate by segment and time period in SQL - Build a cohort retention table - Engineer churn-predictive features (recency, tenure, engagement) - Identify which product features correlate with higher/lower churn - Present findings to a stakeholder with clear recommendations