Quick spatial outliers detecting with random sampling

Existing Density-based outlier detecting approaches must calculate neighborhood of every object, which operation is quite time-consuming.. The grid-based approaches can detect clusters or outliers with high efficiency, but the approaches have their deficiencies. We proposed new spatial outliers detecting approach with random sampling. This method adsorbs the thought of grid-based approach and extends density-based approach to quickly remove clustering points, and then identify outliers. It is quicker than the approaches based on neighborhood queries and has higher precision. The experimental results show that our approach outperforms existing methods based on neighborhood query.