The Application Of Systematic Clustering Analysis In Proficiency Testing

Nowadays, laboratory accreditation activities develop rapidly in society, economic and trade fields, which plays a significant role in promoting economic development of the society. Proficiency Testing as the major content of laboratory accreditation is of great concern in laboratory inspection level, proficiency calibration and so on, and it has been utilized by international organizations as the most frequently used technical approach to inspect proficiency of laboratory. Consequently, more and more laboratory management apartments, laboratory accreditation bodies all over the world and the related international organizations pay much attention to it.Proficiency Testing is the activity of assuring testing proficiency of laboratory by comparisons between laboratories. Putting it in a simple way, Proficiency Testing involves many laboratories inspecting samples of the same quality or characteristic according to the formulated laboratory method, gathering the results, analyzing in a scientific way and then finalizing the judgment of satisfactory or stray according to the deviation degree from the median of each laboratory.The most frequently used way of Proficiency Testing is robust statistical method, however, it lacks in a general judgment and evaluation and the assumption of this method is that the data complies to normal distribution, yet the actual found is that when testing the normal distribution of quantities of data, quite a portion of it doesn’t comply to the assumption. So the "Z" score determined from robust statistical analysis conflicts with the results both from the standard of testing, which makes the analysis of participants of laboratory impossible and results in the inability of determining if the testing results are accurate. As a result, the judges and experts can not evaluate the Laboratory Proficiency Testing efficiently. Hence, if excluding the distribution types of laboratory data, Proficiency Testing is the issue of distributing, to some extent. The following essay applies Group Average Method and Ward’s method to statistical analysis of Proficiency Testing and analyzes each data tested by laboratory by clustering, later compares the results with ones determined from robust statistical method and it is found the two methods of clustering handles the issue of distribution of Proficiency Testing in a efficient way.The context is divided into four chapters:Chapter One:Introduction of the background, definition, significance and international and domestic statuses of Proficiency Testing study.Chapter Two:Introduction of statistical method of statistics used in the essay.Chapter Three:It concerns with statistical analysis of the data determined from each laboratory by robust statistical method、group average method and ward’s method.Chapter Four:Evaluating the results handled in Chapter Three and the problem and prospect of Proficiency Testing.