Identifying customer buying behaviour in medium-scale supermarkets using data mining techniques

Abstract

Data mining with the assist of the most acceptable pattern returning and the frequent algorithm can majorly affect marketing and sales. Frequent pattern (FP) recognition is a considerably researched sector in data mining because of its significance in many day-to-day applications. Market basket analysis is one of its usages; applied by retailers / groceries to discover customers’ buying behaviour from stores. The output of the analysis may increase the profitability of the retail owners, service quality, avoid empty shelves, and customer satisfaction. The research aims to focus on descriptive analysis of the customer buying patterns, items buying together, and item units that are highly bought from the supermarket and find the most efficient algorithm for this task. It facilitates reordering set the supermarket shelves layout. It was performed by identifying the current information to discover and analyse frequent itemsets to illustrate an association rule. Mainly, FP growth and apriori algorithms were used for market basket analysis. Both two algorithms are implemented using python by feeding pre-processed raw data. One algorithm was selected considering the efficiency. Since the FP growth algorithm consumes more execution time against the steps of the algorithm, the apriori algorithm is used to implement the recommendation model. The model which is implemented using the apriori algorithm, was developed to recommend the products using month-wise association rules