An innovative algorithm for clustering retail items

Dendrogram for clustering retail items is a tool used in customer relationship management (CRM) for implementing homogeneous schemes for all the items in one cluster. In agglomerative hierarchical clustering, dendrograms are developed based on the concept of `distance´ between the entities or, groups of entities. These entities may be the customers, retail items, business units etc. as per the business problem. The present paper focuses on agglomerative hierarchical clustering of retail items based on the concept of `similarity´ of the items which is regarded as the inverse of `distance´ (or `dissimilarity´). Similarity is measured based on the strength of association rule/s containing the candidate items to be merged. An algorithm has been proposed for developing dendrogram and it has been explained with an example in retail sale. The algorithm overcomes the negative impact of negative association between the items on the quality of clustering.