Local outlier index-based method to evaluate soil metal contamination

This paper proposed a local outlier index method to evaluate soil metal contamination. The method calculates local outlier index of samples. The bigger the index is, the more the object has a probability of becoming the contaminated one. Additionally, deep outlier and extensive outlier were introduced to ensure the accuracy of experimental results; a data-filtering method which is based on density-sampling was developed to reduce the algorithm complexity when dealing with massive data, by filtering intensively distributed data and reserving the data of spares regions. On real metal contamination data of farmland soils in suburbs of Beijing, the results of the local outlier index algorithm, which is consistent with that of Nemerow index method, demonstrate its effectiveness of analysis of metal contamination problem.