Retail Concept 1: Market Basket Analysis...a brief introduction

While doing your monthly grocery over the week end in any of the retail outlets like Spencer or Big Bazar, you may have noticed that products being located in combinations. So there are cola bottles placed just beside a rack of snacks and also as you arrive at the billing counter you could find a number of small items starting from chewing gums to tobacco packs just beside the billing counter.

 All this a result of the Market Basket analysis, a concept from data mining which helps retailers find the group of products which sell the most in the company of one another. It is the analysis of a set of products which sell together and is also known as affinity grouping.

The basic idea behind Market Basket analysis is simple; analyze your data and find out the combination of products that sell together the most together. You can then place those related items side by side thus increasing the probability of one item being sold when the customer buys the other as he need not do the search.

 There are three basic measurements which guide the market basket analysis. If we are trying the affinity of two products A and B then we have:
  • Frequency
  • Support 
  • Confidence
Frequency: Frequency is an absolute number and gives the total number of orders which contained both products A and B together. For example a store processed 857 order in a day which contained both the orders A and B.
Frequency as such being an absolute number is not of much meaning to the merchandiser until he gets to know Support and Confidence.

 Support:Now if the store as mentioned above processed 1000 orders in a day then the Support for the proposition that both Product A and B would be sold together is 857/1000 or 85.7%..

Confidence:If now lets say that there 900 orders in a day at the store which contained Product A and out of that there were 857 orders in which the customer ordered Product B along with Product A. So, the Confidence factor would be 857/900 or 95.2%.

All the three factors when studied for different combination of products sold in the store would give the retailer a fair amount of idea which are the products which sell the most together and could be grouped together for higher sales.

The term Market Basket might have come from Amazon's shopping basket where if the customer chooses a particular item, a number of other suggestive items are also shown to him which he might be interested in buying.

One of the famous applications of the Market basket analysis was in Walmart where its was noticed that there was a strong association between a brand of babies diapers and a brand of beer.
Analysis of purchases revealed that they were made by men, on Friday evenings mainly between 6pm and 7pm.
After some serious thinking, the supermarket figured out the rationale:
  • Because packs of diapers are very large, the wife, who in most cases made the household purchases, left the diaper purchase to her husband
  • Being the end of the working week, the husband and father also wanted to get some beer in for the weekend
Walmart took advantage of this and place a brand of premium beer is a rack just beside the diapers and beer sales sky rocketed.
There are a number of real life examples where the concept of Market Basket analysis is applied in some derived format or the other, such as:
  • Analysis of credit card purchases.
  • Analysis of telephone calling patterns.
  • Identification of fraudulent medical insurance claims. 
So next time you for a grocery at Big Bazar and Spencer just watch out for more applications of the Market Basket analysis.