Density-Distance Outlier Detection Algorithm Based on Natural Neighborhood

Outlier detection is of great significance in the domain data mining. Its task to find those target points that are not identical most object generation mechanisms. The existing algorithms mainly divided into density-based and distance-based algorithms. However, both approaches have some drawbacks. former struggles handle low-density modes, while latter cannot detect local outliers. Moreover, outlier algorithm very sensitive parameter settings. This paper proposes a new two-parameter (TPOD) algorithm. method proposed this does need manually define number neighbors, introduction relative distance can also solve problem low density further accurately combinatorial optimization problem. Firstly, natural neighbors iteratively calculated, then calculated by adaptive kernel estimation. Secondly, computed through neighbors. Finally, these two parameters combined obtain factor. eliminates influence require users determine outliers themselves, namely, top-n effect. Two synthetic datasets 17 real were used test effectiveness method; comparison with another five provided. AUC value F1 score on multiple higher than other algorithms, indicating be found accurately, which proves effective.