Chapter 8. Association Rule Mining

Amir is the VP of Sales at Big Bonanza Warehouse. The other evening while shopping for cookies on Amazon he got a little message. “People who ordered cookies also ordered cookie-holders.” “Cookie-holders? That’s ridiculous.” He thought. But he clicked on the item anyway. “Cookie-holders are only a buck, I’ll try one.” A moment later he realized, “I bought something I didn’t intend to buy. I’m happy with the purchase and the recommendation. How can I do this for my own sales and customers?”

The next day in the office he called in Duane, the SAP business analyst for Sales. He explained what he was thinking and wanted to know how they could do it. “I want to provide sales recommendations for all my retail locations. When a customer buys a product, I want the system to provide recommendations for related products.” Duane’s first thought was, “SAP doesn’t do that.”

Upon talking to Greg and Paul, Duane learns that what Amir wants can be achieved by using a technique called association rule mining. We intend to take sales orders from SAP and create associations, or discover the general rules of patterns in item purchases. We want to know what products are most often purchased together. Consider groceries: if a customer buys bread and eggs, what is the likelihood they will buy milk?

However, if you understand that association rule mining employs the rules of probability, you start to see many more applications:

Laboratory studies
What is the probability ...

Get Practical Data Science with SAP now with the O’Reilly learning platform.

O’Reilly members experience live online training, plus books, videos, and digital content from nearly 200 publishers.