矿产资源统计回归预测的数据模型

UATA MODELS FOR MINERAL RESOURCES PREDICTION BY STATISTICRAL REGRESSION

根据当前矿产资源定量预测使用线性回归存在的问题,本文探讨了使用二步正态转换获得最佳预测的方法。结论是:如果正态转换能够成功,则符合正态条件的因变量与自变量的数据模型,其预测结果最优;如果正态转换不成功,则选取因变量与自变量在分布参数上最为接近的数据模型亦可使预测结果最优。

On the basis of the problems existing in using linear regression in statistical prediction of mineral resources, an optimal prediction method obtained by means of two-step normal transformation is cxamined. Conclusions show that if normal transformation can succeed and if the data modal of the unpendent and inpendent variables is suitable to normal condition, the forecast results of linear regression are optimum, and that if normal transformation cannot scceed and if the data modal of the distribution parameters of unpendent and inpendent variables that maximally approach is chosen, the forecast results will also be optimum.