An Improved Cuckoo Search Based K-Means for Outlier Detection

To enhance the precision and timeliness of outlier detection techniques under data mining background, the characteristic of clustering algorithm is study and analyzed. Addressing the inherent issue of clustering randomly assigned initial points of K-means algorithm, this study introduces an Improved cuckoo search algorithm base on k-means to detect outlier data. Firstly, the cuckoo search algorithm is improved by changing the constant of discovery probability and Lévy flight step size from a constant to a changing value. Then, the improved cuckoo search algorithm and the K-means are combined into a new outlier detection algorithm. Finally, it is applied to the outlier detection of signaling data. Findings from our experiments indicate that this proposed approach outperforms other methods.