Customer Segmentation using KMeans and PCA

This project performs customer segmentation based on online transaction data using clustering techniques, dimensionality reduction, and visualization.

Dataset

The dataset used is int_online_tx.csv, which contains transaction records including:

• CustomerID
• InvoiceNo
• StockCode
• Quantity
• UnitPrice

Objectives

• Clean and preprocess the transaction data.
• Engineer features representing customer behavior.
• Standardize the data for modeling.
• Apply PCA to reduce dimensionality.
• Perform clustering using KMeans.
• Visualize the customer segments and key statistics.

Steps

1. Data Cleaning: Remove missing CustomerID values and create a Sales column.
2. Feature Engineering: Aggregate features like total transactions, products bought, total sales, and quantities.
3. Standardization: Normalize features using StandardScaler.
4. Dimensionality Reduction: Apply PCA to reduce to 2 principal components.
5. Clustering: Use KMeans with 4 clusters.
6. Visualization:
   • PCA Scatterplot of clusters.
   • Boxplot of total sales by cluster.
   • Count plot of customers in each cluster.
   • Correlation heatmap of engineered features.

Requirements

• Python 3.x
• pandas
• numpy
• matplotlib
• seaborn
• scikit-learn

Install requirements using:

pip install -r requirements.txt