Minimum threshold determination method based on dataset characteristics in association rule mining

Abstract Association rule mining is a technique that widely used in data mining. This to identify interesting relationships between sets of items dataset and predict associative behavior for new data. Before the formed, it must be determined advance which will involved or called frequent itemset. In this step, threshold eliminate excluded itemset also known as minimum support. Furthermore, provides an important role determining number rules generated. However, setting wrong leads failure association obtain rules. Currently, user determines support value randomly. challenge becomes worse ignorant characteristics. It causes lot memory time consumption. because formation process repeated until finds desired The adaptive model based on average total each transaction, well their values. proposed method uses certain criteria thresholds, therefore, resulting are accordance with needs. obtained from utility divided by existing transactions. Experiments were carried out 8 specific datasets determine using different trial 2 basic algorithms rule, namely Apriori Fpgrowth. test repeatedly highest lowest result showed 6 produced maximum values apriori fpgrowth algorithms. means has ability generate when viewed quality produces at lift ratio > 1. characteristics experimental results can factor value.