测量数据中异常值判别方法探讨
Exploration of outliers identification methods in measurement data
侯建明1
作者信息
- 1. 山西省测绘地理信息院,山西 太原 030001
- 折叠
摘要
针对在测量过程中由于测量环境的变化以及人为因素可能会导致个别测量数据值偏离预期或偏离大量统计数据值结果的情况,为了得到客观、科学的测量结果,需要对测量结果中是否存在异常值进行判别及剔除.首先列举了对异常值判别的几种常用方法,然后结合实际案例对测量数据中是否存在异常值,异常值如何判别以及剔除进行了探讨.结果表明:1)如果把异常值数据和正常数据放在一起进行统计,势必会影响试验结果的正确性;2)如果把疑似异常的数据值简单地剔除,又可能忽略了重要的试验信息,可见对测量结果中异常值的正确判别及剔除至关重要.
Abstract
In view of the situation that individual data values may deviate from expectations or deviate from a large number of statistical data values due to changes in external conditions and human factors in the measurement process,in order to obtain objective and scientific measurement results,it is necessary to distinguish and eliminate whether there are abnormal values in the measurement results.Firstly,several common methods for determining outliers are listed,and then combined with practical cases,whether there are outliers existing in the measured data,how they can be distinguished and eliminated are discussed in this paper.The results show that:1)If we put the outlier data and the normal data together,the correctness of the experimental results will be bound to be affected;2)If the suspected abnormal data values are simply eliminated,important experimental information may be ignored.So it is very important to correctly distinguish and eliminate whether there are outliers in the measurement results.
关键词
测量数据/异常值/判别/剔除Key words
measurement data/outlier/distinguish/eliminate引用本文复制引用
出版年
2024
NETL
NSTL
万方数据