使用不同权函数形式对带异方差的思茅松树高曲线方程进行加权回归,筛选出方程的最优权函数形式。根据前人的研究成果,文章提出非定式权函数理论,并以非定式权函数对不同异方差性的分组数据开展研究。结果证明,模型中异方差性的强弱可以用方差线性斜率表示,方程中异方差性的强弱会影响权函数的选择。

This paper used different weighting function to regress the tree height-diameter curve equation of Pinus kesiya var.langbianensis with heteroscedasticity,and selected the optimal weighting function form.According to previous research results,the article proposed a set of variable-weighting function theory,and researched on different heteroscedasticity of group data by variable-weighting function.Study results show that the heteroscedasticity strength in model can be expressed by variance linear slope,and the heteroscedasticity strength in equation will affect the choice of weighting function.