![]() | Only 14 pages are availabe for public view |
Abstract Outlier detection is the process of finding data objects with behaviors that are very different from expectation (outliers). Detecting outliers in a large data set is a major data mining task. The existing approaches in this field are categorized into two main categories: distance-based and density-based outlier detection approaches. Although, Local Outlier Factor (LOF) is considered as the most popular density-based algorithm, it still has some problems related to the speed and accuracy. Enhancing LOF algorithm is the focus of many researchers working in this field. Among the improved versions of LOF, GridLOF has been introduced to have a good performance. This thesis presents a new proposed method to enhance GridLOF algorithm by replacing one of its steps with a less complex step which reduces the complexity to be only ( ) instead of ( ) in a novel way. This will greatly decrease the execution time especially for large data sizes. The simulation results show that the proposed algorithm outperforms GridLOF algorithm in terms of speed while achieving the same accuracy of GridLOF algorithm which is considered as the standard accuracy. |