Customer Segmentation & Market Basket Analysis

GitHub

The Challenge

In the fast-paced and competitive landscape of online retail, businesses encounter numerous challenges when it comes to comprehending and satisfying their customers' diverse needs.

The analysis focuses on two crucial aspects to address these challenges:

  1. Conducting cluster analysis to segment customers based on their purchasing behavior.
  2. Performing market basket analysis to reveal product associations within transactions.
By harnessing the combined potential of cluster analysis and market basket analysis, the ultimate aim is to gain comprehensive insights into customer behavior, elevate business decision-making, and unlock strategic growth opportunities.

The Context

The analysis was done with The Online Retail II dataset, sourced from UC Irvine's Machine Learning Repository, illuminates the world of customer behavior in e-commerce. Spanning from 01/12/2009 to 09/12/2011, it unveils a fascinating journey:

  1. Transaction Footprint: With 1,067,371 entries across 8 columns, it's a window into online shopping dynamics.
  2. Key Columns: InvoiceNo for unique transaction IDs, StockCode for product identification, Description for product details, Quantity for purchased amounts, InvoiceDate for transaction timestamps, UnitPrice for item costs, CustomerID for unique customer IDs, and Country for geographical context.

The Findings

The analysis provided valuable insights into customer behavior, market dynamics, and revenue sources for the business. It unveiled a deeper understanding of customer behavior throughout the years, months, weekdays, and hours.

Geographically, the UK emerged as the dominant market, followed by Germany and France. To capitalize on this, businesses should continue nurturing customer relationships in the UK while strategically targeting Germany and France for further growth.

The months of October, November, and December stood out as the months with the highest transaction percentages, indicating peak sales during the holiday season.

To leverage this seasonal trend, companies must prepare by adjusting inventory levels, optimizing marketing campaigns, and ensuring sufficient resources to meet customer demand.

Personalize & Engage: Client Clustering

Four customer segments were identified, each with unique strategies to engage and retain them:

  1. The New Customer: These customers have not spent much but show potential for future engagement. Strategies to keep them engaged include introducing them to new products, offering discounts or deals, implementing a loyalty program, and sending targeted marketing campaigns.
  2. The Loyalist: These customers are already engaged with the company frequently and are valuable to the business. Strategies for this cluster involve enhancing the customer experience, personalizing offers and recommendations, rewarding their loyalty, and seeking feedback.
  3. The Forgotten / Unattended: These customers have made a few purchases but are not engaged with the company. Strategies to re-engage them include targeted campaigns, offering incentives, and implementing a win-back program.
  4. The Irregular: These customers have spent significantly on specific products but are not regular shoppers. Strategies to encourage repeat purchases include cross-selling and upselling, creating product bundles or promotions, sending reminders or notifications, and collecting feedback.

Unveiling Shopping Patterns: Market Basket Analysis

The analysis also uncovered that customers tend to buy complementary products together or sets of different colors.
Recommendations based on these findings include cross-selling, bundling and discounts, personalized recommendations, and targeted marketing campaigns.

The Process

Using R Studio, the analysis was done as followed:

Step 1: Importing and Understanding the Data

The dataset was imported into RStudio, and a thorough understanding of the variables was achieved.

Step 2: Tidying and Transforming the Data

The data was cleaned, formatted, and enriched with additional attributes, ensuring data integrity for meaningful insights.

Step 3: Unleashing the Power of Models

The K-means algorithm was used for customer segmentation, and a market basket analysis was performed to reveal distinct customer groups and product associations.

Step 4: Visualizing Insights

Captivating visualizations using ggplot, radar charts, and grouped matrix-based visuals were employed to effectively communicate the findings.

Customer segmentation and market basket analysis are powerful tools that provide invaluable insights into customer behavior and preferences in the online retail landscape. By adopting data-driven strategies based on these insights, businesses can optimize their operations, enhance customer engagement, and drive sustainable growth in the highly competitive market.

Source: Chen, Daqing. (2019). Online Retail II. UCI Machine Learning Repository. https://doi.org/10.24432/C5CG6D.