Tree Based Space Partition of Trajectory Pattern Mining for Frequent Item Sets
P. Geetha, E. Ramaraj · SSRN Electronic Journal · 2016
Pattern discovery from the large data base is an interdisciplinary field in a computer science termed data mining. The prediction of spatial trajectory locations of these patterns is an attractive research area leads to the evolution of frequent mining algorithms. The existence of the diverse databases degrades the execution time of the mining process. To overcome this issue, Tree based Space Partition of Trajectory Pattern Mining (TSPTPM) is proposed in this paper. A novel method, which adopts clustering and tree structure to extract the frequent patterns from the real time diverse datasets. Initially, the clustering process organizes the number of transactions into group and assigns the ID to each group. Then, heap tree algorithm constructs a tree from the cluster of transactions. In a tree structure, the transaction with maximum ID consider as the root node and the odd and even number of transactions consider as child nodes. Then, pre-order traversal extracts the frequent patterns in the tree structure. Finally, Vague Space Partition (VSP) algorithm applied to offer the flexible spatial partition of tree structure and convert into sequences. The proposed TSPTPM enhances the frequent pattern extraction process by reducing time consumption. The optimized patterns are obtained from the heap tree structure and the obtained results are effectively partitioned by using the novel VSP algorithm. The mining time analysis for various datasets namely, Chess, Mushroom, Connect and accident with various number of items in cluster. The comparative analysis of mining time for proposed TSPTPM algorithm and existing Apriori proves the effectiveness.