稳健统计学与地球化学数据的统计分析

ROBUST STATISTICS AND GEOCHEMICAL DATA ANALYSIS

Due to the existance of various errors and outliers , observed geochemical data often deviate from the normal or lognormal distribution and thus are not suitable for classical statistical methods . Robust statistics provides methods which give reliable results even when the data deviate to a certain extent from statistical assumptions . This paper introduces basic ideas of robust statistics . The necessity of adapting the robust estimation of location and scale as well as robust regression for estimating geochemical background and delineating geochemical anomalies , and the necessity of adapting robust principal components analysis for studying the correlations between geochemical variates are expounded . Examples of the application of robust methods in geochemical data analysis are provided .