Evaluation Of Outlier Detection For Trajectory Data

Outlier of trajectory dataset is different from other in this trajectory dataset. The outlier is involved according to human error sensors or mechanical faults and system behavior or environment. It becomes challenges in accuracy of clustering classification and other data mining task. The problem statement is how to detect the outlier and what will be more effective method to detect these outliers. Outlier is detected to increase data quality for all applications. To detect these outliers similarity measurement is used. In this system Longest Common Subsequence LCSS based measurement and Housdroff Distance HD are applied. A comparison result of experimental study on effectiveness of these two methods is described. Experimental observation demonstrates that LCSS Distance produces better results than the other algorithm.