📊 Telecom Customer Churn Analysis

Exploratory Data Analysis & Feature Engineering

📌 Project Overview

Customer churn is a critical challenge in the telecom industry. This project focuses on analyzing customer behavior and identifying key factors influencing churn through detailed Exploratory Data Analysis (EDA) and robust Feature Engineering techniques.

🎯 Objectives

• Understand customer demographics and service usage patterns
• Identify key drivers of customer churn
• Perform feature extraction and feature selection to enhance model performance

📂 Dataset

Dataset available on the Kaggle platform Telecom Customer Churn dataset. This dataset includes a variety of variables that represent customer profiles and behaviors, forming the basis for predicting the likelihood of churn.

🔍 Exploratory Data Analysis (EDA) + Data Visualization

• Analysis of churn distribution across customer segments
• Visualization of relationships between churn and key variables
• Detection of trends and correlations

⚙️ Data Preprocessing

• Encoding categorical variables
• Scaling numerical features

🛠 Feature Engineering

🔹 Feature Extraction

Several new features were created to better capture customer behavior:

• Tenure_in_years: Converted customer tenure from months to years for easier interpretation of subscription duration.
• ChargesRatio: Ratio of Total Charges to Monthly Charges, providing insight into how long a customer has been using the service based on spending patterns.
• LevelTotalCharges: Categorized Total Charges into spending levels (Low, Medium, High) to segment customers based on overall spending.
• LevelTenure: Grouped customer tenure into categories such as New, Mid-Term, and Long-Term to simplify analysis of subscription length.

🔹 Feature Selection

• Correlation analysis to remove redundant features
• Statistical tests to evaluate feature significance
• Reduced dimensionality while preserving predictive power

📈 Key Insights

• Long-term contracts significantly reduce churn risk
• Customers with higher monthly charges and shorter tenure are more likely to churn
• Value-added services play a critical role in customer retention

🚀 Outcome

• A refined dataset ready for machine learning models
• Clear understanding of churn-driving factors
• Actionable insights to support customer retention strategies

🧰 Tools & Technologies

• Python
• Pandas, NumPy
• Matplotlib, Seaborn
• Scikit-learn

📎 Next Steps

• Build and evaluate machine learning models for churn prediction
• Perform hyperparameter tuning and model optimization
• Deploy insights through dashboards or APIs