论文部分内容阅读
矿产资源预测中常用的矿床模型法有两大要素:一是反映矿床地质成矿规律的数据;二是符合这一数据特点的统计预测法即优良的回归法。这就是由常用的线性回归法和分布求同法组成的求同回归法。由于分布求同法是一种符合线性回归先进数学前提(即相同分布论)的数据预处理,又有现实的可能性,所以求同回归法不仅可以克服传统线性回归法和其他回归法无法克服的非线性局限性,而且还可以克服稳健回归在自变量中有异值(outliers)时识别和消除异点影响不力的弊病,从而使这种新的求同回归法在各种回归中处于最优的地位。最后,文中给出了两个很好的利用该法进行铀资源定量预测的实例。
There are two main elements of the mineral deposit model commonly used in mineral resource prediction: one is the data that reflects the geological mineralization law of the deposit; the other is the statistical regression method that accords with the characteristics of this data, that is, the excellent regression method. This is by the commonly used linear regression and distribution of the same law composed of the same law of regression. Since the law of distribution seeking is a kind of data preprocessing which accords with the advanced mathematics premise of linear regression (ie the same distribution theory), and there is a realistic possibility, seeking the same regression method can not only overcome the traditional linear regression and other regression methods can not overcome But also can overcome the shortcomings of robust regression in recognizing and eliminating outliers when there are outliers in the independent variables so as to make this new symbiosis regression method the most important among all kinds of regression Excellent position. Finally, two examples of the quantitative prediction of uranium resources using this method are given.