Comparison of Apriori and Frequent Pattern Growth Algorithm in Predicting The Sales of Goods

Abstract

The increasing number of bona fide companies, especially in the world of retail minimarkets, PT. Suka Maju innovates to make a company that develops in the retail sector so that it can serve consumers well. With the problems - problems in the company PT. Suka Maju still applies unrelated items so that consumers find it difficult to buy related products. PT. Suka Maju does not apply interrelated items such as coffee and sugar, sauce and noodles, bread and cheese. company PT. Suka Maju must act as quickly as possible and requires data analysis using Market Basket Analysis. The purpose of the existence of data in every transaction of product sales to consumers, data can be processed properly to provide information to companies so that transaction data in every product purchase can be useful and to determine the layout of a product. To deal with this problem, researchers found a pattern that can improve a layout pattern or display of sales items in the retail world, one of which is by utilizing product sales transaction data used to support and find an association rule data mining method technique, comparing the algorithm Apriori and algorithm Frequent Pattern Growth. The purpose of this study is to compare 2 algorithms and choose a better algorithm to help find products that are often purchased together. From the results of the research from 10,005 transactions of 27 attributes using the algorithms Apriori and algorithms Frequent Pattern Growth with the minimum parameters of support = 100, confidence = 100 and lift = 2.58, the algorithm Frequent Pattern Growth has the highest accuracy compared to the algorithm Apriori. In the results of this study, it can be said that the algorithm Frequent Pattern Growth is the best for determining interrelated