Sequential Approach for Mining of Temporal Itemsets

Abstract

Sequential approach for mining temporal itemsets initially proposed by Yoo and Sekhar uses the Euclidean distance measure to discover similarity profiled temporal associations. This paper extends the sequential approach proposed by Yoo by applying the proposed dissimilarity function. The proposed dissimilarity measure for sequential method is obtained by extending the basic Gaussian membership function. Experiments are conducted by applying naïve and sequential approaches using Lp-norm distance function over synthetic dataset generated and results are compared to the sequential approach using proposed Gaussian based distance function. Naïve and sequential methods using Euclidean distance function and sequential approach using proposed distance function are compared w.r.t computational time and computational space. Experiment results using synthetic datasets proved that the performance of proposed approach is better to naïve and sequential approaches in terms of computational time.