论文部分内容阅读
本文探索了半变异函数基台值与样品方差的关系。潜在变异函数基台值等于总体方差真值,这一点虽然早已公开发表,但在处理实际数据时,无论是潜在的变异函数还是总体方差的真值都是未知的。据此,本文对把样品方差用作变异函数基台值的估计值的做法提出疑义,并为确定这种做法的合法性提出了一个简单的概念性构思。发现了一个更加有用得多的结论,即样品方差的期望值等于所有可利用的样品数据对间半变异函数值的算术平均值。这个结论不要求作任何平稳性的标准假设:它只需假设有变异函数存在。
This article explores the relationship between the semivariogram base station values and sample variance. The latent variation function is equal to the base value of the population variance. Although this has been published for a long time, both the potential variation function and the true value of the population variance are unknown when dealing with actual data. Based on this, this paper raises doubts about the use of sample variance as an estimate of the base value of the variability function, and proposes a simple conceptual idea for determining the legitimacy of this approach. A much more useful conclusion was found that the expected value of the sample variance was equal to the arithmetic average of the values of all available sample data pairs. This conclusion does not require any standard hypothesis of stationarity: it simply assumes that a variogram exists.