Mining Frequent Itemsets Using Proposed Top-Down Approach Based on Linear Prefix Tree (TD-LP-Growth)

Plenty of algorithms are available for datamining. LP-Growth occupies an important place in data mining. LP-Growth algorithm constricts data required for mining frequent itemsets in LP-tree and recursively builds LP-tree to mine entire frequent itemsets. In this study, an algorithm of top-down linear prefix tree (TD-LP-Growth) is proposed for mining frequent itemsets. The proposed TD-LP-Growth algorithm searches LP-tree from top to down order which is opposite to the old LP-Growth algorithm. TD-LP-Growth does not generate conditional pattern base and conditional LP-tree. Thus, it improves the performance of proposed TD-LP-Growth algorithm. In this paper, the benchmark databases considered are Online shopping dataset 1, Chess and Mushroom. While using online shopping dataset, the frequent purchaser of the dataset is visualized using Google map in geographical method. From the experimental results, it is concluded that the proposed TD-LP-Growth algorithm consumes lower runtime and memory space during the process of mining. Thus, the proposed 1D-LP-Growth algorithm outperforms LP-Growth algorithm in mining frequent itemsets.