<![CDATA[The growth of online business has been rising considerably in recent years. The growth is affected by technology advancement in Internet and smartphones and consumer behavior change for better online shopping experience. To anticipate this swift customer behavior, business owners need to have an excellent inventory management to be able to keep making profits. In data mining realm, the algorithm model that is known to be applied in this case is the association algorithm. This model will explicate customers’ purchasing patterns where is useful in calculating stock accurately. The aim of this research is to find an appropriate model in handling large data to obtain valid association rules that have minimum support value, confidence value, and high lift ratios. It is hoped that the results of this research can provide recommendations for online sellers to manage a large variety of goods and to keep making profits. Datasets that contain a large variety of goods are handled first by using a clustering algorithm to group similar items together. The dataset tested was divided into three groups, namely, dataset without clusters, k-means cluster, and agglomerative cluster. After forming three groups of datasets, FP-Growth was applied to each dataset. The result is that datasets with clusters, whether using k-means or agglomerative, have a minimum support value that is greater than datasets without clusters. Most association rules are obtained from the k-means cluster dataset. Based on the model applied in this research, the association itemset size only obtains one conclusion from one premise.]]>
