A case study of customer segmentation and product association in agricultural business

Nowadays, competition in agricultural business has been much more intense. Market strategic planning has become a major key for increasing sales volume. A good strategy must be appropriate for handling different desires of customers. In this project, we aim to segment customers into several groups that share similar behaviors. Then we analyze customer behaviors in each group and propose strategic guidelines for strategic planning according to customers’ behaviors. The transaction data used in this work was collected by a case study agricultural firm for 24 months consisting of 573 business customers. For customer segmentation, we apply K-means clustering with elbow method, NC correlation, NC index I and II used for finalizing the number of groups. For products association, we apply association rule mining via apriori algorithm in a few selected groups from the K-means clustering.

The results show that the customers are segmented into 8 groups. The three main groups are “regular customers”, “loyalty customers” and “irregular customers” which include 26.71, 25.13 and 23.21 percent of all customers, respectively. Then, we apply the association rule mining to the "loyalty customers " and "regular customers" to analyze the association of products. By setting the minimum support of 1% and minimum confidence of 40%, the algorithm yields 537 and 24 rules from the loyalty and regular groups, respectively. In this work, we focus on two interesting types of the rules: (A→B) and (C→D) where A is a set of necessary products, B is a supplementary product, and C and D are products to be used together. By convincing the customers that only ordered either A or C to order B or D may lead to an increased sales volume for the firm. Finally, the case study firm may apply the results of customer segmentation and the analysis of customers’ purchasing patterns as a strategic guideline for marketing planning that appropriately fits customers and boosts its sales volume.