HomeInsightseCommerce Data Prep with AWS DataBrew and Glue for Personalized Recommendations

eCommerce Data Prep with AWS DataBrew and Glue for Personalized Recommendations

Updated 1 Apr 2026

Intuz eCommerce Team

The Data Transformation Challenge

Imagine you have an e-commerce web site with tens of millions of items and hundreds of millions of user activity. How do you turn uncooked, unstructured data into smart, contextual recommendations? That's where magic comes in.

AWS DataBrew: Your Data Transformation Wizard

Honestly say data is rarely perfect right out to gate. It’s more like a rough diamonds that must need polishing. So, here AWS DataBrew is secret weapon for this data transformation.

Key DataBrew Transformations

1. Product Rating Aggregation

We didn't want to just look at individual scores – we wanted to know the real quality of a product. So, we applied a group-by transformation to find the average score per product. This provides us with a more objective perspective:

• Group all scores by product ID
• Calculate average score
• Make a product_rating_mean that captures general product quality

Why is this significant? Because one upset customer or one over-the-top reviewer shouldn't destroy or make a product.

2. Price Cleaning

Ever try to do math with prices that have currency symbols? Nightmare. We fixed this by:

• Eliminating those annoying ₹ symbols
• Converting text prices to real numbers
• So we can actually do mathematical calculations on them

3. Smart Categorization

We didn't just want to see raw prices. We set up sensible price buckets:

• Budget: Less than ₹300
• Mid-range: ₹300-₹500
• Premium: More than ₹500

This gives us insight not only into the price, but the market positioning of products.

4. Data Integration

We created several transformed files in DataBrew. But how do we get them to get along? Meet AWS Glue and Athena.

AWS Glue Crawler: The Automatic Organizer

Imagine the Glue Crawler as an über-intelligent librarian. It:

• Reads all our transformed data in S3

• Automatically discovers schemas and types

• Fills AWS Glue Data Catalog without having to do anything

AWS Athena: Simplified Querying

Athena allows us to query our refactored data like a regular database. We utilized it to:

• Validate data quality
• Establish a consistent picture of all our revamped datasets
• Allow our recommendation engine to be built on a clean, stable data base

Why This Matters

In online shopping, personalization isn't just a bonus — it's essential. Even a small improvement (like 1%) in product recommendations can lead to a big increase in sales.

When we work with data — like rearranging, mixing, or rotating it — we're not just playing with numbers. We're doing it to:

• Understand how people behave,
• Predict what they might like next,
• And create those surprising moments where the site seems to "just know" what the customer wants.

Ready to turn your raw e-commerce data into sales-driving recommendations?

Talk to our experts to see how AWS DataBrew and Glue can transform your customer experience and boost sales with personalized recommendations.