ASSOCIATION RULE MINING ON LIBRARY BOOKS LENDING DATA USING APRIORI AND JACCARD SIMILARITY

Abstract

UPT Perpustakaan UNS has 37,271 collections and on average 75,316 annual circulations of the book that is managed by UNSLA (UNS Library Automation). An analysis is needed to discover valuable information that can be used for various purposes. Association rule mining is one of data mining techniques to look for relationship pattern in the market basket data. Apriori algorithm is commonly used in association rule mining. However, Apriori has limitations in conducting association rule mining on sparse data. Jaccard Similarity algorithm is used to find the similarities between the two sets. Application of Jaccard Similarity to the association rule mining can find association rule on sparse data. This research was conducted to determine the consistency of association rule generated by the combination of both Apriori and Jaccard Similarity compared to regular Apriori and Jaccard Similarity on the book lending data of UPT Library UNS. Data are grouped into ten different categories of books and split by month and year. Association rule mining is done by using all three methods. Association rules produced by each method compared for consistency in the known month and year. As a result, it is known that the association rule mining using a combination of Apriori and Jaccard Similarity is more consistent than the original Apriori and Jaccard Similarity. However, association rule mining using Jaccard Similarity generate more variation than Apriori and combination. UPT Perpustakaan UNS has 37,271 collections and on average 75,316annual circulations of the book that is managed by UNSLA (UNS LibraryAutomation). An analysis is needed to discover valuable information that can beused for various purposes. Association rule mining is one of data mining techniquesto look for relationship pattern in the market basket data. Apriori algorithm iscommonly used in association rule mining. However, Apriori has limitations inconducting association rule mining on sparse data. Jaccard Similarity algorithm isused to find the similarities between the two sets. Application of Jaccard Similarityto the association rule mining can find association rule on sparse data. This researchwas conducted to determine the consistency of association rule generated by thecombination of both Apriori and Jaccard Similarity compared to regular Aprioriand Jaccard Similarity on the book lending data of UPT Library UNS. Data aregrouped into ten different categories of books and split by month and year.Association rule mining is done by using all three methods. Association rulesproduced by each method compared for consistency in the known month and year.As a result, it is known that the association rule mining using a combination ofApriori and Jaccard Similarity is more consistent than the original Apriori andJaccard Similarity. However, association rule mining using Jaccard Similaritygenerate more variation than Apriori and combination.