📊 Customer Segmentation Analysis – Final Report
- Project Description
The goal of this project is to perform customer segmentation for an e-commerce company using the marketing_campaign.csv dataset. By analyzing demographics, spending behavior, and campaign responses, customers are grouped into segments. These insights help improve targeted marketing strategies, customer satisfaction, and business performance.
- Dataset Overview
Rows: 2,240 customers
Columns: 29 features (demographics, spending, purchases, campaigns)
Key Features:
Demographics: Year_Birth, Education, Marital_Status, Income, Dt_Customer
Spending: MntWines, MntFruits, MntMeatProducts, … MntGoldProds
Purchases: NumWebPurchases, NumCatalogPurchases, NumStorePurchases
Campaigns: AcceptedCmp1–5, Response
Others: Recency, Complain
- Data Cleaning & Feature Engineering
Missing values in Income handled using median.
Converted Dt_Customer to datetime (format = %d-%m-%Y).
Created new features:
Age = 2025 – Year_Birth
Total_Spent = Sum of all product spending columns
Dropped irrelevant columns: ID, Z_CostContact, Z_Revenue.
- Exploratory Data Analysis (EDA)
Age Distribution: Most customers are 30–60 years old.
Income vs Spending: Positive relationship (higher income → higher spending).
Campaign Response: Most customers did not respond; response rate is low.
(Include plots: Age histogram, Income vs Spending scatter, Campaign Response countplot).
- Customer Segmentation (K-Means Clustering)
Features used: Income, Recency, Total_Spent, NumWebPurchases, NumCatalogPurchases, NumStorePurchases.
Standardized data and applied KMeans (k=4).
Cluster Profiles:
Cluster Avg Income Avg Age Avg Spending Recency Insights 0 High 40s High Recent buyers Premium segment – target with luxury campaigns 1 Medium 30s Medium Recent buyers Growing customers – nurture loyalty 2 Low 50s Low Older customers Value-seekers – discounts work best 3 Medium 60s Low-Mid Not recent Dormant – need re-engagement strategies
(Include scatterplot: Income vs Total Spending colored by cluster).
- Insights & Recommendations
High spenders: Focus on premium products, exclusive offers.
Value-seekers: Provide discounts, bundled offers.
Dormant customers: Send reactivation campaigns (emails, coupons).
Young professionals: Target digital/online campaigns, as they shop more online.
- Learning Outcomes
Gained practical experience in clustering algorithms (KMeans).
Improved data cleaning & feature engineering skills.
Learned visualization techniques to present insights.