🧠 Customer Segmentation Using K-Means Clustering

This project applies unsupervised machine learning to segment customers based on their demographics, spending habits, and purchasing behavior.
It also includes a Streamlit web application that predicts a customer’s segment using a trained K-Means model.

---

🚀 Live Deployment

🔗 Live Application Link:
👉 https://customer-segmentation-mlproject.streamlit.app/

---

🚀 Features

• Segments customers into 6 meaningful groups
• Performs data cleaning, feature engineering, and EDA
• Uses Elbow Method to determine the optimal number of clusters
• Visualizes clusters using PCA scatter plot
• Saves trained model and scaler as .pkl files
• Streamlit app for real-time cluster prediction

---

🛠 Technologies Used

• Python
• Pandas, NumPy
• Matplotlib, Seaborn
• Scikit-Learn (KMeans, StandardScaler, PCA)
• Joblib (Model Saving)
• Streamlit (Web App)

---

📂 Files Included

• Analysis_Model.ipynb → Data analysis, feature engineering, clustering, PCA visualization
• app.py → Streamlit app to predict customer segment
• customer_segmentation.csv → Dataset used
• kmeans_customer_segmentation_model.pkl → Trained KMeans model
• scaler_customer_segmentation.pkl → Saved StandardScaler
• MLPPT_CustomerSegmentation.pdf → Project PPT
• MLProjectReport_CustomerSegmentation.pdf → Full project report

---

📊 Project Workflow

1️⃣ Data Preprocessing

• Removed missing values
• Converted date columns
• Cleaned and formatted dataset

2️⃣ Feature Engineering

Created new features:

• Age
• Total Spending
• Total Children
• Customer Since (days)

3️⃣ Exploratory Data Analysis

Visualized:

• Age distribution
• Income distribution
• Total spending
• Boxplots
• Correlation heatmap

4️⃣ K-Means Clustering

• Selected 7 important features
• Standardized data using StandardScaler
• Determined optimal k = 6 using the Elbow Method
• Evaluated clusters using Silhouette Score

5️⃣ PCA Visualization

• Reduced data to 2 components
• Displayed cluster separation using a PCA scatter plot

6️⃣ Model Saving

Saved using joblib:

• kmeans_customer_segmentation_model.pkl
• scaler_customer_segmentation.pkl

---

🌐 Running the Streamlit App

▶ Install dependencies

pip install -r requirements.txt ▶ Run the application streamlit run app.py

▶ App Functionality

User enters:

Age

Income

Total Spending

Web Purchases

Store Purchases

Web Visits

Recency

The app displays: 👉 Predicted Customer Segment (Cluster 0 – 5)

📁 Project Structure ML_CustomerSegmentation/

├── Analysis_Model.ipynb

├── segmentation.py

├── customer_segmentation.csv

├── kmeans_customer_segmentation_model.pkl

├── scaler_customer_segmentation.pkl

├── README.md

├── MLPPT_CustomerSegmentation.pdf

└── MLProjectReport_CustomerSegmentation.pdf

🎯 Results

Optimal number of clusters: 6

PCA scatter plot shows clear separation of clusters

Identified meaningful customer groups for targeted marketing

Fully functional Streamlit prediction application

👩‍💻 Author

Durga Sravanthi Peddoju

B.Tech CSE

SRM University AP