Advanced Customer Segmentation

This project performs advanced customer segmentation using clustering techniques on a dataset of mall customers. The app leverages KMeans clustering and visualizations to uncover patterns in customer data and provide insights for targeted marketing or strategic business decisions.

Features

• File Upload: Users can upload a CSV file containing customer data.
• Data Preprocessing: Handles missing values and removes outliers.
• Clustering: Uses KMeans clustering to segment customers based on selected features.
• Optimal Clusters Detection: Uses the Silhouette Score and the Elbow Method to suggest the optimal number of clusters.
• Visualizations:
  • PCA-based 2D & 3D Cluster Visualizations
  • Cluster Size Pie Chart
  • Correlation Heatmap
  • Cluster Profiles (Boxplots & Radar Charts)
  • Dendrogram for Hierarchical Clustering
  • Pairwise Feature Distribution
  • Distance to Centroid Distribution
• Cluster Quality Metrics: Computes and displays the Silhouette Score for clustering quality.
• Cluster Metrics: Displays mean values for selected features within each cluster.

Requirements

• Python 3.7+
• Streamlit
• Pandas
• Numpy
• Plotly
• Seaborn
• Matplotlib
• Scikit-learn
• SciPy

You can install all required packages using the requirements.txt file:

pip install -r requirements.txt

Usage

1. Run the app with Streamlit:

   streamlit run app.py

2. Upload your CSV file containing customer data.

3. Select features to cluster on, and the app will display visualizations and metrics based on the clustering.

Example Data Format

The dataset should include columns such as:

• CustomerID: Unique identifier for each customer
• Gender: Customer's gender
• Age: Customer's age
• Annual Income (k$): Customer's annual income in thousands
• Spending Score (1-100): Customer's spending score (1 to 100)

Example:

CustomerID,Gender,Age,Annual Income (k$),Spending Score (1-100)
1,Male,19,15,39
2,Male,21,15,81
...

Acknowledgments

• The project uses KMeans clustering from scikit-learn for customer segmentation.
• Plotly, Matplotlib, and Seaborn are used for creating interactive and static visualizations.
• The hierarchical clustering dendrogram is generated using SciPy's linkage function.

---

This README provides a basic overview and usage instructions for the project. Feel free to modify or expand it based on additional details you'd like to include.