Comparison of hybrid approaches with traditional algorithms for improving scalability of frequent pattern mining

Frequent pattern mining always occupies its space in research activities in spite of various emerging research ideas. This paper focuses on performance of four competitive algorithms namely hybrid k-Direct Count and Intersect algorithm with apriori algorithm, hybrid k-Direct Count and Intersect algorithm with transaction mapping algorithm, Modified Ant Colony algorithm and improved Ant Colony algorithm over frequent pattern mining against traditional algorithms. We focus on a set of well defined parameters such as database layout, scanning of input databases, memory requirement, input/output cost, input/output overhead, computational cost, execution speed, scalability, support for parallelization, complexity of the algorithm to do this comparison. Experimental results support for this effective comparison with help of visualization through graphs. We used ASP.net as front end tool and SQL server 2005 as back end tool for implementing our proposed approaches. As a result of this study, there is good improvement in effectiveness of pattern mining in all aspects while using the improved Ant Colony algorithm against other hybrid approaches and traditional algorithms.