Method for judging doubtful values of sample detection data

Abstract

The invention relates to a method for judging doubtful values of sample detection data. The method comprises the following steps of: A, respectively carrying out n tests on detection data of same elements of same samples in P laboratories; B, calculating an intragroup repeatability value r; C, if the range of n numerical values of the ith laboratory is greater than or equal to the repeatability value r, retesting or abandoning the laboratory, and otherwise, indicating that the laboratory passes repeatability judgement; D, repeating the steps B to C, and re-calculating the repeatability value r until the residual laboratories pass the repeatability judgement; E, if the range of the average value of the laboratories which pass the repeatability judgement is greater than or equal to a reproducibility value R, retesting or abandoning the test numerical values of two corresponding laboratories; and F, considering the residual test numerical values as qualified numerical values. According to the method provided by the invention, the dead zones of common Grubbs test and Dixon doubtful value test are overcome, more scientific judgement is carried out on data during constant value analysis, and beneficial supplementation is carried out on the traditional guide sample development data statistical analysis method.

Description

technical field [0001] The invention relates to a method for analyzing standard sample data, in particular to a method for judging suspicious values ​​of sample detection data. Background technique [0002] Usually, the methods and rules of mathematical statistics are used in the statistical analysis of sample detection data, such as Grubbs test, Dixon test, etc. There are often blind spots in the Sidney test and Dixon test. In the case of non-normal distribution (that is, there are extreme values), some abnormal data cannot be found and eliminated, which reduces the accuracy and reliability of the detection data. The Chinese invention patent "Method for Judging Suspicious Value of Sample Test Data" (ZL201410088717.6) is based on Robust statistics, using the median value and standard interquartile range to analyze and judge the test data. In addition, other judging methods can also be established to judge the data from different angles when analyzing the fixed value. Cont...

AI Technical Summary

Problems solved by technology

[0002] Usually, the methods and rules of mathematical statistics are used in the statistical analysis of sample detection data, such as Grubbs test, Dixon test, etc. There are often blind spots in the Sidney test and the Dixon test. In the case of non-normal distribution (that is, there are extreme values), some abnormal data cannot be found and eliminated, which reduces the accuracy and reliability of the test data.

Images