16-05-2012, 03:35 PM
Market Basket Analysis in a Multiple Store and Departmental Environment
OBJECTIVE:-
Market basket analysis (also known as association-rule mining) is a useful method of discovering customer purchasing patterns by extracting associations or co-occurrences from stores’ transactional databases. Because the information obtained from the analysis can be used in forming marketing, sales, service, and operation strategies, it has drawn increased research interest. The existing methods, however, may fail to discover important purchasing patterns in a multi-store environment, because of an implicit assumption that products under consideration are on shelf all the time across all stores. In this paper, we propose a new method to overcome this weakness. Our empirical evaluation shows that the proposed method is computationally efficient, and that it has advantage over the traditional method when stores are diverse in size, product mix changes rapidly over time, and larger numbers of stores and periods are considered.
EXISTING TECHNOLOGY & ISSUES:-
By far, the Apriori algorithm [1] is the most known algorithm for mining the association rules from a transactional database, which satisfy the minimum support and confidence levels specified by users. There are two main problems in using this algorithm in a multi-store environment. The first is caused by the temporal nature of purchasing patterns. An apparent example is seasonal products. Temporal rules [3] are developed to overcome the weakness of the static association rules that either find patterns at a point of time or implicitly assume the patterns stay the same over time and across stores. A literature review on temporal rules is given by Roddick and Spiliopoulo. In temporal rules, selling periods are considered in computing the support value, where the selling period of a product is defined as the time between its first and last appearances in the transaction records. Furthermore, the common selling period of the products in a product set is used as the base in computing the ‘‘temporal support’’ of the product set. The results of the method may be biased, however, because a product may be on shelf before its first transaction and/or after the last transaction occurs, and a product may also be put on-shelf and taken off-shelf multiple times during the data collection period.
The second problem is associated with finding common association patterns in subsets of stores. Similar to the problem in using existing temporal rules in a multi-store environment, we have to consider the possibility that some products may not be sold in some stores, for example, because of geographical, environmental, or political reasons. This is seemingly related to spatial association rules. However, the focus of spatial rules is on finding the association patterns that are related to topological or distance information in, for example, maps, remote sensing or medical imaging data and VLSI chip layout [2].
OUR GOAL:-
To overcome the problem, we propose a method, called store-chain association rules, specifically for a multi-store environment, where stores may have different product-mix strategies that can be adjusted over time.
The format of the rules is similar to that of the traditional rules. However, the rules also contain information on store (location) and time where the rules hold. The rules extracted by the proposed method may be applicable to the entire chain without time restriction, but may also be store and time specific.
The proposed rules have a distinct advantage over the traditional ones because they contain store (location) and time information so that they can be used not only for general or local marketing strategies (depending on the results), but also for product procurement, inventory, and distribution strategies for the entire store chain.
An Apriori-like algorithm is developed for mining chain-store association rules. A simulation is used to empirically compare the proposed and traditional association-rule mining methods.
Three factors are considered in generating stores’ sales data:
(1) the numbers of stores and periods,
(2) the store size, and
(3) the product replacement ratio.
The analysis of the simulation result suggests that the proposed method has advantages over the traditional method especially when the numbers of stores and periods are large, stores are diverse in size, and product mix changes rapidly over time.
Store-chain association rules represent a promising research area in data mining. The results of this paper can be extended by considering time constraints, spatial constraints, quantitative attributes and/or taxonomy, and other kinds of time- or location-related knowledge.
Furthermore, it is important to explore the strategies of generating the store-chain association rules incrementally, in an on-line model, in a distributed environment, or in parallel models.
HOW IT WORKS:-
To overcome the problems, we develop an Apriori-like algorithm for automatically extracting association rules in a multi-store environment. The format of the rules is similar to that of the traditional rules.
However, the rules also contain information on store (location) and time where the rules hold. The results of the proposed method may contain rules that are applicable to the entire chain without time restriction or to a subset of stores in specific time intervals.
An example is: ‘‘In January, customers purchase cold medicine, humidifiers, coffee, and sunglasses together in supermarkets near skiing resorts.’’ These rules can be used not only for general or localized marketing strategies, but also for product procurement, inventory, and distribution strategies for the entire store chain.
Furthermore, we allow an item to have multiple selling time periods; i.e., an item may be put on-shelf and taken off-shelf multiple times. We further assume that different stores can have different product-mixes in different time periods.
That is, each store can have its own product- mix, and the product-mix in a store can be dynamically changed over time. Because the time and store (location) factors are considered, the rule generation procedure is more complicated than the Apriori algorithm.
The proposed method is computationally efficient and has significant advantage over the traditional association method when the stores under consideration are diverse in size and have product mixes that change rapidly over time.