Neighborhood outlier detection

KNN (k nearest neighbor) is widely discussed and applied in pattern recognition and data mining, however, as a similar outlier detection method using local information for mining a new outlier, neighborhood outlier detection, few literatures are reported on. In this paper, we introduce neighborhood model as a uniform framework to understand and implement neighborhood outlier detection. Furthermore, a neighborhood-based outlier detection algorithm is also given. This algorithm integrates rough set granular technique with outlier detecting. We propose a neighborhood-based metric on outlier detection, and compare neighborhood outlier detection with DIS, KNN and RNN. The experimental results show that neighborhood-based metric is able to measure the local information for outlier detection. The detected accuracies based on neighborhood outlier detection are superior to DIS, KNN for mixed dataset, and a litter better than RNN for discrete dataset.