An improved and efficient frequent pattern mining approach to discover frequent patterns among important attributes in large data set using IA-TJ-FGTT

Data mining involves discovering interesting patterns from large dataset to maximize the profit of the future business. Association rule mining is the main area in the field of data mining exploration with wide range of applications. Determining the frequent item-sets in large dataset is the core task of association rule mining and it is frequently used by business decision makers to improve their future business strategy. Numerous algorithms have been dedicated in the literature and tremendous progresses have been made to find frequent patterns. Most of the methods introduced in the literature find all frequent patters for all attributes in the dataset. We introduced TR-FCTM (Transaction Reduction-Frequent Count Table method) for the same. Even though this method outperformed than Apriori and FP-tree, the performance of that technique was reduced slowly when the total attributes in the data bank increases. Sometimes it is needed to catch all significant patterns for a few significant related attributes selected from total attributes in the data bank by the field expert to improve the business in future. So this IA-TJ-FGTT (Important Attributes-Transaction Joining-Frequency Gathering Table Technique) is proposed and its performance is compared with FP-tree. Experimental results of IA-TJ-FGTT shows that this technique outperforms than FP-tree.

I. Introduction

After the invention of internet, most of the business organizations collect and save large amounts of data from day to day operations as more and more works are computerized [1]. Any organization in the world has begun to understand that the data collected in the past is the significant intentional benefit and it also ascertains that the prospective intelligences are secreted in the large volume of data. So it needs techniques to mine the most prized information from large dataset [2].