ASSOCIATION ANALYSIS

ASSOCIATION ANALYSIS

1. • Market Basket Analysis

   • Where the buying habits of customers are discovered through searching for items that are frequently purchased together
     • This can be useful for marketing managers in planning promotions or defining the range and location of products on the shelf.

2. • Web Mining

   • To understand the frequency with which combinations of web pages are visited by a given user in a single session or several sessions.
     • Useful for web analysts in improving the navigation between web pages and
       • also placing advertisement and other promotional banners on web pages.

3. • Credit Card Purchases

   • Useful for credit card companies to know which purchases are associated with payment by credit cards
     • In this way, they can direct future credit card promotions for such purchases.

4. • Fraud Detection

   • The association of certain patterns or incidences in transactions may flag off a potential fraud which may justify investigation by the company
     • EG: a purchase requisition submitted by staff A is approved by staff B, who is a relative to staff A.

5. ADVANTAGES

   • Association rules
     • simplicity and ease of interpretation.
   • When using association rule for prediction
     • AA can be extended to include more sophisticated applications
       • EG
         • A particular item can be identified as a target variable and association rules that have that item as the consequent can be generated.
           • This enables decisions to be made with respect to that item (e.g., promotion or discount).
           • Also, in the case of market basket analysis, in addition to items being included in the analysis to understand which ones are usually purchased together, customer characteristics (e.g., age and gender) can also be included.
             • This enables the organisation to study, for instance, the relationship between purchasing patterns and demographic patterns.
               • Further, instead of examining items that are purchased, it is also possible to include items that are not purchased.
                 • In this case, an associate rule may be: if peanuts and potato chips are purchased and cola is not purchased, there is a 65% chance that beer is also purchased.
   • Time-incorporated AA
     • examine the sequence in which the items are purchased.
       • A sequence association rule EG.
         • if paint and brushes are purchased at time t (e.g., current visit),
           • then there is a 75% chance that drills will be purchased at time t+1 (e.g., next visit),
         • t = time period
       • applications such as the study of repair sequences or web navigation sequences
         • to increase on-line purchases, an on-line bookstore may try to isolate the sequences of web navigation that are likely to lead to on-line purchases. Actions (e.g., web re-design) can then be taken to promote sales-generating sequences/behaviour.

6. LIMITATIONS