Interview and Opinion
Now Reading
Market Basket Analysis and Association Rules

by Sapience Consultancy

“IF … THEN …” is a conditional statement that is broadly used in programming languages. The same idea is also used in analyzing purchase behavior and finding patterns. Which is based on the theory that if you buy a certain group of items it is more (or less) likely to buy another group of items.

Maybe you were waiting at the counter in some grocery store and bought a pack of chewing gum or a candy bar. Or perhaps on a trip to a convenience store to purchase a soap you also picked up a shampoo and a conditioner. It is not coincident that gums and candies or shampoos and conditioners are stocked near each other. Usually these types of recommendation systems and the ways items are stocked are based on subjective experience of marketing professionals and inventory managers.

With the advancement of data mining and machine learning, techniques could be used on large amount of transnational data to uncover hidden and interesting patterns in purchasing behavior. Market Basket Analysis is a method used for finding associations in large databases by using statistical performance measures. The information revealed from Market Basket Analysis help retailers to manage better categories by understanding buyers’ needs and developing cross-promotional programs and occasion-based consumption insights, or even capture new buyers.

The result of Market Basket Analysis is a set of association rules that specify patters of relationships between items. A typical rule might be expressed in the form: IF {Eggs, Butter, Baking Powder} THEN {Balloons}, this is type of rule that is based on consumption occasion which indicates that buyers usually buy these items when they are preparing a cake for a birthday party.

Although association rules are used in market basket analysis and supermarket shoppers, they are also very useful in finding patterns in many other types of data. Like analysis of credit card purchases, identifying fraud medical insurance claims, recommending books in online stores based on previous preference, recommending movies or series based on past viewing behavior etc.

Sapience Consultancy has been recently working on association rules and algorithms that are strong in tackling large amount of transnational data and producing results that are easy to understand.

For more details please contact us on

To learn more about Sapience Consultancy, please visit our website at

About The Author
Insight World Editor