🚀 Customer Segmentation using Clustering Algorithms

---

📌 Project Overview

This project focuses on unsupervised learning techniques to perform customer segmentation using airline data.
Multiple clustering algorithms such as K-Means and DBSCAN are implemented and compared to identify meaningful customer groups.

The goal is to uncover hidden patterns in customer behavior for better business decision-making.

---

❗ Problem Statement

Businesses struggle to identify different customer types from raw data.

This project solves that by grouping customers into meaningful segments using clustering.

---

🎯 Objectives

• Perform clustering on airline customer data
• Identify optimal number of clusters using Elbow Method
• Apply and compare:
  • K-Means Clustering
  • DBSCAN Clustering
• Visualize customer segmentation
• Derive actionable business insights

---

📊 Dataset

• Dataset: EastWest Airlines Dataset
• Contains customer-related features such as:
  • Balance
  • Bonus Miles
  • Flight Miles
  • Transactions
  • Customer activity metrics

---

⚙️ Tech Stack

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

---

🔍 Methodology

1. Data Preprocessing

• Handled missing values
• Feature selection
• Data normalization using scaling techniques

2. Exploratory Data Analysis (EDA)

• Distribution analysis
• Correlation checks
• Pattern identification

3. K-Means Clustering

• Applied K-Means algorithm
• Used Elbow Method to determine optimal clusters
• Segmented customers into distinct groups

4. DBSCAN Clustering

• Implemented density-based clustering
• Identified noise and outliers
• Compared performance with K-Means

---

📈 Results & Insights

• Successfully segmented customers into meaningful clusters
• Identified different customer behavior patterns
• DBSCAN helped detect outliers and noise points
• K-Means provided clear and interpretable clusters

👉 These insights can help businesses in:

• Targeted marketing
• Customer retention strategies
• Personalized services

---

💡 Key Learnings

• Importance of feature scaling in clustering algorithms
• Differences between partition-based (K-Means) and density-based (DBSCAN) clustering
• How to evaluate clustering performance using visual techniques like the Elbow Method
• Handling real-world unsupervised datasets and identifying patterns
• Detecting noise and outliers using DBSCAN

---

🧠 Business Interpretation

• Cluster 1: High-value frequent flyers → Target with premium offers
• Cluster 2: Low activity customers → Re-engagement campaigns
• Cluster 3: High bonus usage → Loyalty-focused strategies

---

🔮 Future Enhancements

• Apply Hierarchical Clustering
• Use PCA for dimensionality reduction
• Deploy as an interactive app using Streamlit
• Automate cluster interpretation using AI

📸 Visualizations

🔹 Elbow Method

🔹 Cluster Visualization

---

📊 Evaluation

• Elbow Method used to determine optimal clusters
• DBSCAN used to identify noise and outliers

---

🚀 How to Run the Project

git clone https://github.com/MeghanaCVarghese/Customer-Segmentation-Clustering.git
cd Customer-Segmentation-Clustering
pip install -r requirements.txt

## Run the notebook:
jupyter notebook notebooks/clustering_analysis.ipynb

---

## 👩‍💻 Author

Meghana C Varghese
Data Scientist | Machine Learning Enthusiast