Finding Suitable Threshold for Support in Apriori Algorithm Using Statistical Measures

Abstract

Data mining is one of the noticeable researches. Data mining is one of the most outstanding areas of research, striving to extract meaningful information and useful models from databases and to help managers to make better decisions. Mining association rules is a persistent and popular data mining process. It aims to find frequent models, association correlations, or causal structures between a set of objects in large transactional or relational databases and other data repositories. This paper provides an improvement of the Apriori algorithm, a classic rule extraction algorithm by finding appropriate minimum threshold values for the support automatically using different statistical measures depending on each dataset. The experiments on baseline datasets show a comparative analysis between different proposed methods of generation association rules using the automatic minimum support threshold.