Understanding association rules
Video tutorial
Association rules find patterns in very large amounts of point-of-sale data. An association rule is of the form “If a shopper purchases Item A and Item B, the shopper also purchases Item C.” For example, association rules might indicate that if a shopper buys onions and potatoes on a trip to the supermarket, she is likely also to buy hamburger meat. Such information can be used as the basis for decisions about marketing activities, such as promotional pricing and product placements.
To identify association rules, you must provide the transaction ID and the transaction items for a very large number of purchases. The Transaction Id and Transaction Item columns must belong to the same database table. You must also provide values for Minimum Support and Minimum Confidence:
*
Support (A,B) = Transactions (A,B)/Total transactions
*
Confidence (A,B-> C) = Support (A,B,C)/Support (A,B)
Use the default values for Minimum Support and Minimum Confidence for the first trial. Depending on the results, you can increase or decrease these values.
Optional parameters include:
*
*
*
*
*
The results of a trial are displayed in a table. The antecedent and consequent items appear in the first two columns. There are also several numeric columns that measure how well the rule predicts the consequent. The columns that appear in the Results tab are:
*
*
*
*
*
*
*
Items in the antecedent and consequent are enclosed in double quotation marks and separated by a comma. Numeric rule measures such as support, confidence, and lift are rounded to two decimal places.
How to find association rules
1  
2  
3  
Figure 6-17  
4  
Figure 6-18  
5  
Figure 6-19  
6  
Choose Calculate. The association rules appear in the Results tab, as shown in Figure 6-20. Decreasing the values for Minimum Support and Minimum Confidence yields several opportunities, indicated by a gold star.
Figure 6-20  
Video tutorial
Using Association Rules
 

Additional Links:

Copyright Actuate Corporation 2013 BIRT Analytics 4.2