Influence of Discrete Granularity on Using Association Rules to Complete Missing Values

This paper investigates the impacts of different discrete granularities of continuous attribute on the algorithms which employ association rules to complete missing value. The Chi2 algorithm is used to discrete the continuous attributes. By using three different discrete granularities, the impacts of the algorithm on completing missing value are explored. Experimental results show that different discrete granularity has negligible impacts on using association rules to complete missing value. However, the completing accuracy has an inclination to decrease with the decrease of the discrete granularity.