Discarding Abnormal Measurements

The article explores estimation issues of mixed sample and measurements rejection. The article reveals the content of statistical parametric estimates. The estimation limitation of the maximum likelihood on finite samples is shown. The difference in asymptotic estimate and the estimate for finite sample is shown. The meaning of robust estimates is explained. The situation of measurements is considered in the presence of measured quantities two sets, one of which is unknown for measurements subject. The concept of pure sample and mixed sample is introduced. A mixed sample is formed when an element of the second set falls into dimensions sample of the first set which distorts the statistical estimates. The element of the second set in the sample of the first set is uncountable or abnormal. The article proposes a new method for eliminating uncountable measurements. The method increases the analysis and processing of the sample by n-times #CSOC1120.