Exploration of Trustworthy Association Rules for Market Basket Analysis

Market basket analysis (MBA) is an effective strategy for the retailers to increase product selling. Earlier concepts of MBA relate with statistical analysis of data. However, statistical analysis remains silent in exploration of customer purchase pattern. Knowledge of customer purchase patterns is necessary to design effective selling strategies such as promotion of product combination, discount on combo package etc. Association rule mining (ARM) explores the knowledge of customer purchase patterns from the basket data. Traditional support and confidence-based ARM techniques often generate a large volume of rules including spurious ones. Decision making with spurious rules often leads to promotion of loss-making products and demotion of profit-making products. This fact thus compromises trustworthiness of the association rules. This paper proposes a new approach of trustworthy association rule mining (TARM) to generate quality association rules from the transactional database. The proposed approach designs a new parameter called soundness in association of associability, gain and accuracy of the rules. The outcomes of the experiments with real datasets show satisfactory performance of the proposed approach.