Mall Customers Clustering

This repository contains a clustering analysis of the Mall Customers dataset. The main work is in code/clustering_project.ipynb, where the data is explored, preprocessed, and segmented using unsupervised learning.

Dataset

The project uses dataset/Mall_Customers.csv, which contains 200 mall customers with the following attributes:

• CustomerID
• Genre
• Age
• Annual Income (k$)
• Spending Score (1-100)

The dataset has no missing values.

Goal

The objective is to identify meaningful customer groups based on income and spending behavior, then compare two clustering approaches:

• K-Means clustering
• Agglomerative hierarchical clustering

The notebook uses Annual Income (k$) and Spending Score (1-100) as the main features for clustering after scaling.

Workflow

The analysis follows these steps:

1. Load and inspect the dataset.
2. Explore feature distributions and pairwise relationships.
3. Remove CustomerID, which does not help with behavioral clustering.
4. Encode Genre and scale the numeric features.
5. Estimate the best number of clusters with inertia and silhouette analysis.
6. Fit K-Means with the selected K.
7. Compare K-Means with hierarchical clustering.
8. Visualize the final segments and summarize the customer profiles.

Main Result

The notebook selects K = 5 as the optimal number of clusters. The final groups are interpreted as customer segments such as average shoppers, premium shoppers, impulsive shoppers, careful spenders, and sensible shoppers.

Repository Structure

• code/ - notebooks for the clustering analysis
• dataset/ - Mall Customers CSV file
• paper/ - space for report material

Requirements

The notebook was developed with Python and common data science libraries, including:

• pandas
• numpy
• matplotlib
• seaborn
• scikit-learn
• scipy

How To Run

1. Open code/clustering_project.ipynb in Jupyter or VS Code.
2. Make sure the relative dataset path points to dataset/Mall_Customers.csv.
3. Run the notebook cells from top to bottom.

Reference

The notebook also includes a comparison against scikit-learn clustering documentation and a Kaggle reference for the dataset (https://www.kaggle.com/datasets/abdallahwagih/mall-customers-segmentation/data).