Abstract—Various methods for mining association rules at multiple conceptual levels focusing on different sets of data and applying different thresholds at different levels have been proposed in literature. These are ML_T2L1, ML_T1LA, ML_TML1, and ML_T2LA. It has been observed that these algorithms show higher processing time and processing cost as well as need large amount of memory space. This paper focuses on the comparative performance evaluation of the ML_TMLA algorithm that generates multiple transaction tables for all levels in one database scan with that of ML_T2L1 and ML_T1LA algorithms. The performance study has been conducted on different kinds of data distributions (three synthetic and one real dataset) and thresholds, which identify the conditions for algorithm selection. The Tool used for the experimental and comparative evaluation of the proposed algorithm with other algorithms is the AR Tool. It has been concluded that the ML_TMLA algorithm performs better than all the algorithms mentioned above.

Index Terms—Data mining, Knowledge discovery in databases, Association rules, multiple-level association rules

Cite: Suraj Srivastava, Harsh K. Verma, and Deepti Gupta, "On Performance Evaluation of Mining Algorithm for Multiple-Level Association Rules based on Scale-up Characteristics," Journal of Advances in Information Technology, Vol. 2, No. 4, pp. 234-238, November, 2011.doi:10.4304/jait.2.4.234-238