Fuzzy Clustering Approach to Consumer Behavior Analysis Based on Purchasing Patterns

Abstract

Consumer behaviour analysis is critically important to contemporary marketing strategy, allowing for an understanding of purchasing patterns and design of targeted interventions. Traditional segmentation methods like K-means and hierarchical clustering are often not sufficient as they force consumers into a single cluster, though life is not set in stone and real worldbehaviour is continuously overlapping. This research utilizes fuzzy clustering (machine learning) to segment consumers across Product Category Preference (PCP), Price Sensitivity (PS) and Purchase Frequency (PF) using the Fuzzy C-Means (FCM) algorithm. The report identifies two underlying consumer segments: High Spend Customers, who have strong product preference, purchase often, and are not sensitive to price, and Bargain Shoppers who are price-sensitive and make infrequent purchases with low product preference. Fuzzy clustering assigns each customer as a member to each of the segments with a value between 0 and 1, rather than in a traditional, discrete method of forcing segmentation, capturing the flexible, context-dependent nature of purchasing behaviour. In this process, the model iterates over the computation of cluster centroids and updates cluster membership until it reaches a stable state. The study finds that fuzzy clustering is more representative of the hybrid and uncertain nature of consumer behaviour which provides businesses with the benefit of adapting marketing strategies accordingly. Beyond the segmentation of consumers, fuzzy clustering may also find application in personalization, recommendation engines (including recommendation and personalization of products), dynamic pricing (where customers are targeted on special offers), loyalty programs, and optimization of the retail space (to better serve customers using both traditional and modern selling formats).