MBA

Sorry bloggers I am really talking about MBA but not MBA (Master of Business Administration)

I am sure a lot of you like me have used Amazon or at the least visited one of the big supermarkets, you may have or not a shopping list and while you are shopping you will find most of the items you want close to each other. In Amazon it will recommend which items you may be interested in and what the other customers view or purchased beside what you select. So it may induce the customer to buy things may like but doesn’t need in the specific moment. Will this save our time or increase their profit? Well depends on how we want to think about it.

Thanks to the technology, it is called Market Basket Analysis (MBA), one of the applications of data mining, in a simple words it studies the customer purchasing habits then identifies the pattern between a combination of items( and sometimes in what sequence) within the basket then provides cross sell proposition. MBA (or sometimes called Product Affinity Analysis) it is a data mining technique based on a theory that if a customer buy certain group of items (referred by MBA as item-set, the customer is more likely to buy another group of items.

MBA is executed through application software that uses scalable algorithm to process the data and perform operation and eventually provide useful outcomes. Computer collect the data of the transactions (and other related data if available like customer id, specific product, time, location and their demographic information) and through data mining technique called association rule. This based on the rule of IF /THEN; IF the probability that a customer will buy certain thing referred as support, THEN the customer is more likely to buy another thing referred as confidence. They studied the probability that a customer will buy a certain item when s/he buys another item. This can go far, some stores study the relationship not only between the items but between items and the customer’s demographic information. So for each customer the MBA application software will compare all items in a basket with all other items in the other basket e.g. field customer id, product x, product y then count the number of items each combination exists or the period under consideration. And then apply the other factors to understand the more accurate nature of the relationship between the items. (Anbarasi et al, 2011)

It may seem simple to compare the relationship between two specific items in one basket to other basket which is considered as first level basket analysis. But it is more than that, with the growth of business and the large number of products, different stores and globalization. The process of  MBA may be slow or of less significance. Differential basket analysis is the solution to such problem to certain limit which is referred as second level basket analysis. It will produce the desired outcome and thus reduce the potentially high volume of trivial outcomes. The second level analysis is more complicated and advanced, not just comparing   the baskets but also apply the effect of other factors like demographic criteria, different stores, days of weeks and different seasons of the year on that products relationship.(Anbarasi et al, 2011)

Algorithm is the key of data mining in MBA which identify the association rule. The most common algorithms used for performing the MBA are Similis algorithm and Apriori algorithm. Apriori algorithm applies mathematical computation that achieves very good scalability and can handle multiple items in the itemset. Similis algorithm is used for more complex and large set of data, which express the transaction data into a graph based structure and then perform the association rule in a relational table. There are other algorithms like Argwal and Swami that able to compute the support and confidence but very limited to single item .Several software available performing the data mining and most classified according to the procedure of analysis; statistical, machine learning and neural network. Furthermore, many softwares performing the MBA like the Oracle and Net Perception.(Cavique, 2007)

In conclusion, MBA is a successful technology through a software application of data mining with a powerful and positive input to businesses and targeted marketing.

References:

Anbarasi , D. Sathya Srinivas, K. Vivekanandan. Dimension reduction techniques for Market Basket Analysis. IPCSIT vol.6 (2011) IACSIT press, Singapore.

Luis Cavique. Ascalable algorithm for the market basket analysis. Journal of Retailing and Consumer Services,vol.14 (Nov.2007) p.400-7.