摘要
实验数据处理 ,常用相关系数的大小来表示线性拟合优劣的判据 .通过实验例证分析研究 ,相关系数好不一定直线拟合结果就好 ,实验结果线性拟合得好不一定相关系数就很大 ,说明不宜用相关系数作实验结果拟合优度的判据 .只有它为回归系数时 ,或当有一变量不可能用一个固定的标度来测量时 ,才有实际意义的 .
In the experimental data disposal, correlation is often used as Criterion of linear fitting goodness. However, examples from some concrete experiment show that good linear fitting experiment data may not have larger correlation coefficient that Is to say, correlation coefficient is not suitable to be a criterion judging goodness of fitting. Only if it is also a regression one coefficient, orthere is one variable that can t be measured with a fixed scale, the correlation coefficient is of significant.
出处
《郑州工业大学学报》
2001年第3期72-74,共3页
Journal of Zhengzhou University of Technology
基金
河南省自然科学基金资助项目 (994 0 62 3 0 0 )
