| What is Market Basket Analysis? | | | | purchases. By doing this, retailers would be able to |
| It is a known fact that customers come to retail | | | | detect the underlying statistical relationships |
| shop, purchase various products and leave the | | | | between products by looking at how often two |
| retail shop. But, do the retailers know that the | | | | or more items appear in the same context or |
| products being brought have certain kind of | | | | were interacted with during the same shopping |
| relationship to each other and what could be the | | | | session. For Example: |
| impact of this kind of insight on their sales? | | | | IF {Men and Diapers and Friday Evening} THEN |
| Market-basket analysis is an effective way to | | | | {Beer}. |
| ascertain the relationship between products. It is | | | | The probability that men would buy diapers on |
| an analysis made on customers purchasing | | | | Friday evenings is referred to as the support for |
| behaviour during same visit and single order. Data | | | | the rule. The conditional probability that a |
| mining techniques are used to ascertain the | | | | customer will purchase beer is referred to as the |
| combination of categories within a customer's | | | | confidence. |
| "basket". | | | | If the transaction level data (t-log data) or loyalty |
| A retailer may group its customers based on | | | | card data is available. There is a great opportunity |
| purchases and can do various analyses to | | | | for the retailer to increase his profitability by |
| determine which different products should be | | | | understanding the components of their customer's |
| merchandized together to improve sales and | | | | basket accurately. Using data analysis software, |
| revenue. This analysis helps in deciding the store | | | | mathematical techniques and intelligent analysis, |
| layout, product assortment, promotional plans and | | | | high tech retailers can leverage the data to |
| cross selling opportunities. | | | | connect with the customers to increase both |
| One way to perform this analysis is to first sort | | | | customer satisfaction and revenues. |
| the customers into clusters based on their | | | | |