Purchase Pattern Analysis with FP-Growth Algorithm for Product Sales Promotion Recommendation Strategies

Abstract

The development of retail business technology is related to the need for management to meet customer demands by using technology. To help make effective sales strategic decisions, it is necessary to optimize the use of information technology on existing sales transaction data. The transaction database that has been stored as a company archive asset can be used for processing information that is useful in increasing product sales and promotions. This study aims to provide an analysis related to the product sales pattern of PT. X in Sumbawa Besar city. PT. X is a retail company that sells distributes daily consumer goods. The algorithm used is Frequent Pattern – Growth which is one of the algorithms in data mining used to find relationships in large data based on the number of occurrences of these data relationships. The Association Rule Mining method can be used in the retail business field, known as Market Basket Analysis. The application used for testing is Rapidminer 9.10. The research stages include: data collection, data preparation, FP-Growth algorithm implementation, result analysis and conclusions. The results of the tests carried out resulted in 819 rules with a total of 85 rules. The results of grouping strong rules based on the combination and number of products that produce information that is expected to be used as recommendations to promote products with discount, cross-selling, up-selling, product bundling and other types of promotions to increase product sales.