Telecom Customer Churn Analysis

Overview

Customer churn is a major challenge in the telecom industry, impacting revenue and customer retention.
This project utilizes PySpark to analyze customer data, identify key churn factors,
and build predictive models to improve retention strategies.

Tech Stack

• Programming Language: Python
• Big Data Framework: Apache Spark (PySpark)
• Machine Learning: Spark MLlib
• Visualization: Matplotlib, Seaborn
• Data Processing: Spark DataFrames

Dataset Overview

The dataset contains telecom customer details, service usage, and churn status.

• 📌 CustomerID - Unique customer identifier
• 📌 Tenure - Number of months a customer has stayed
• 📌 MonthlyCharges - Monthly bill amount
• 📌 TotalCharges - Total amount paid by the customer
• 📌 Contract - Type of contract (Month-to-Month, One Year, Two Years)
• 📌 Churn - Whether the customer left (Yes/No)

🔍 Exploratory Data Analysis (EDA)

✅ Performed data cleaning and handled missing values using PySpark DataFrames
✅ Identified feature correlations using groupBy() and aggregations
✅ Visualized churn distribution using Seaborn and Matplotlib

🤖 Machine Learning Approach

1️⃣ Feature Engineering

✔️ Converted categorical variables using StringIndexer
✔️ Scaled numerical features using MinMaxScaler

2️⃣ Model Selection (PySpark MLlib)

🚀 Logistic Regression
🚀 Random Forest Classifier
🚀 Gradient Boosted Trees (GBTClassifier)

3️⃣ Evaluation Metrics

📌 Accuracy
📌 Precision & Recall
📌 F1-score
📌 Evaluated models using BinaryClassificationEvaluator

📌 Key Findings

✅ Customers with month-to-month contracts have the highest churn rate
✅ Higher monthly charges increase churn probability
✅ Customers with longer tenure are less likely to churn

📜 License

This project is licensed under the MIT License.