THE OUTLIER SAMPLE EFFECTS ON MULTIVARIATE STATISTICAL DATA PROCESSING IN GEOCHEMICAL STREAM SEDIMENT SURVEY (MOGHANGEGH REGION, NW OF IRAN)

7th International Scientific Conference - SGEM2007, www.sgem.org, SGEM2007 Conference Proceedings/ ISBN: 954-918181-2, June 11-15, 2007

ABSTRACT

In geochemical stream sediment surveys in Moghangegh Region in NW of Iran,
sheet 1:50,000, 152 samples were collected and after analyze and processing of
data, it is revealed that Yb, Sc, Ni, Li, Eu, Cd, Co, As contents in one sample is
far higher than other samples. After detecting this sample as the outlier sample,
affect of this sample was investigated in multivariate statistical data processing
to investigate destructive effects of outlier sample existence in geochemical
exploration. Pearson and spearman correlation coefficient methods and cluster
analysis were used for multivariate studies and Scatter Plot of some elements
together with their regression lines are given in case of 152 and 151 samples
and compare the results. After investigation of multivariate statistical data
processing results, it is revealed that results of existence of outlier samples may
appear as these relations between elements:

1- True relation between two elements, which have no outlier frequency in the
outlier sample.

2- False relation between two elements which one of them has outlier frequency
in the outlier sample.

3- Complete false relation between two elements which both have outlier
frequency in the outlier sample.

Keywords: outlier sample, multivariate statistical, geochemical stream
sediment.