论文部分内容阅读
本文首先论述统计法的哲理基础、应用的必要、主要类别、各类的性质及目前的情势;其次具体地分析现时通用的皮尔逊Ⅲ型频率线的性质及其在应用中的各种缺点。对於一般统计法,以往学者曾指出其中存在着的缺点,并建议了各种补救的方法;本文中一一予以评论,并指出还没有为各国学者所发现的主要缺点的癓结所在。 1954年洪水後,林平一指出从资料中算出的统计特性係数之不可靠性,引起了广泛的注意;於是这一问题才在我国开始受到重视,并在中央水利部组织了讨论会。本文分析了依据资料计算特性係数不可靠的原因,否定了现行统计法中这样计算的方法,并建议新的统计方法。以往很多学者误变为统计法存在的问题的主要癓结在概率分佈律,因而採用了或创造了各种派别的概率线用以统计。本文分析了主要的十一派概率线的性质,认为仍以概括性皮尔逊频率公式或葛蘭-沙炎频率线最为合理而广泛地适用于各种资料的分佈。作者认为:合理地应用统计法的唯一途径是试探特性係数C_V、C_S等各种配合下的概率线和资料绘点间离差的程度,从而定出最接近的概率线的C_V,C_S等。在建议的方法中,採用了图解和试算参合的步骤,可以简便地推算出适线的各係数,以及适线的或然误。.文中以算例说明现行统计法的误差可能达到很大的程度。建议的方法免除了现行从资料中计算C_V,C_S等各种存在着的缺点;创造了条件使可能探用繁复的概括性频率线来统计,因此使整个统计方法的应用较为合理化;这在现行统计法中,由于特性係数不可能从资料算得好,是不可能的。
This article first discusses the philosophical foundation of statistics, the necessity of application, the main categories, the nature of various types, and the current situation; secondly, it analyzes the nature of the current common Pearson III frequency line and its various shortcomings in application. For the general statistics law, past scholars have pointed out that there are shortcomings and suggested various ways to remedy; in this article, one by one to comment, and pointed out that there are no major shortcomings for the scholars found in various countries. After the flood in 1954, Lin Ping-yi pointed out that the unreliability of the statistical characteristic coefficients calculated from the data attracted wide attention; therefore, this issue began to receive attention in China, and a seminar was organized in the Central Ministry of Water Resources. This paper analyzes the reason for the unreliability of the coefficient of calculation based on the data, denies the method of calculation in the current statistical method, and suggests a new statistical method. In the past, many scholars mistakenly converted the problems of statistical methods to the probabilistic distribution law, and thus adopted or created probability lines of various factions for statistics. This paper analyzes the nature of the main 11-point probability lines, and considers that the generalized Pearson frequency formula or the Gülen-Sahmatitis frequency line is the most reasonable and widely applicable to the distribution of various data. The author believes that the only way to apply the statistical method reasonably is to test the degree of dispersion between the probability lines and the plotted points of the data, such as the characteristic coefficients C_V, C_S, etc., to determine the C_V, C_S, etc. of the closest probability line. In the proposed method, the steps of graphic and trial calculation are used, and the coefficients of the line of fit can be easily calculated and the coincident errors can be calculated. In the paper, the examples show that the error of the current statistical method may reach a great degree. The proposed method eliminates the existing drawbacks of calculating C_V, C_S, etc. from the data; creating conditions that make it possible to use complex generalized frequency lines for statistics, thus rationalizing the application of the entire statistical method; this is currently In the statistical method, it is impossible because the characteristic coefficient cannot be calculated from the data well.