文章基本信息

标题：UNSUPERVISED LEARNING AS A DATA SHARING MODEL IN THE FP-GROWTH ALGORITHM IN DETERMINING THE BEST TRANSACTION DATA PATTERN
本地全文：下载
作者：MUSTAKIM ; ULYA KHAIRUNNISA ; ALEX WENDA 等
期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2021
卷号：99
期号：11
语种：English
出版社：Journal of Theoretical and Applied
摘要：Market Basket Analysis is an analysis related to consumers and products in marketing. One of the successes of a company in the retail sector depends on promotion and shopping cart analysis. The data patterns generated from an association-based analysis are mostly applied by companies, one of which is the use of data mining technology. FP-Growth has been known as a reliable algorithm in terms of association, but some obstacles in its implementation in the field are often not finding a rule if using a diverse dataset. Unsupervised Learning or what is often known as grouping techniques such as K-Means, K-Medoid, and Fuzzy C-Means (FCM) can divide optimal data based on euclidean distances so that it finds better data patterns than without data sharing, especially in the case of FP-Growth. Comparisons are made by experimenting with the number of clusters 2 to 7, each of which is applied to the clustering algorithm. The results of these experiments, K-Medoid is the algorithm with the best validity value compared to other algorithms. Besides, the use of unsupervised learning techniques combined with FP-Growth can generate rules for each algorithm compared to simply applying FP-Growth.
关键词：K-Means;K-Medoids;FCM;FP-Growth;Data Sharing;Silhouette Index;Cluster Validity