Analisis Pengembangan Dalam Penerapan Recommender System Menggunakan Metode Algoritma Apriori Dan K-Means Clustering Pada Aplikasi E- Commerce. (Studi Kasus Di Big Sport Tangerang)

Abstract

The sales and marketing system in Big Sport is still carried out conventionally, causing the problem of sales transactions which causes a decrease in turnover. The solution to this problem is an e-commerce application for Big Sport and implementing a strategy recommendation system. By implementing the a priori algorithm method used to find out product recommendations on Big Sport to look for products that frequently appear (frequent itemset) with a minimum support calculation of 3 and a minimum Confidence of 50% from sales transaction data in June 2023 from 18 product data to determine the Association Rule for a combination of itemsets that gets an average lift ratio test value of 1.67 with a maximum Confidence value of 100% which forms 22 Association Rule results to provide good and accurate product recommendations for e-commerce applications based on sales transaction history data . The K-Means Clustering method was implemented using tolls rapidminer using transaction data for 6 months from 18 products. From the rapidminer run, the results from cluster 0 contain 8 items, cluster 1 has 7 items, and cluster 2 has 3 items with an average value. within a centroid distance of 2381.332, where cluster 0 has a value of 1975.234, cluster 1 has a value of 2995.918 and cluster 2 has a value of 2030.222. It can be concluded that items in cluster 0 are products with low sales levels, items in cluster 2 with medium sales levels, and items in cluster 1 with high sales levels. And the Davies Bouldin Index value is 0.462 which shows the fact that the centroid distance assessment results are almost close to 0 which can be concluded to have satisfactory results because the lower the DBI value, the better the cluster value so that it can be used as a reference in product procurement.