企业合作进行数据分析时,很多时候不愿意透漏各自的私有数据.针对这种情况提出了在数据异构分布下通过线性转化和添加随机噪声两种数值转换方法来保护企业的私有数据,阐述了怎样在转换后的数据中得到原始的线性回归方程以及线性回归方程拟合优度的计算,怎样在转换后的数据中对回归方程的线性关系和回归系数进行显著性检验,最后结合算例对这两种数值转换方法进行了分析和对比,实验证明基于线性转化和添加随机噪声的两种数值转换方法可用于解决企业合作中私有数据的线性回归问题.
Enterprises are often reluctant to disclose their private data when they analyze data cooperatively.Two conversion methods of the linear transformation and adding random noises are proposed to protect enterprises private data.The calculation of the original linear regression coefficients and the original goodness-of-fit in transformed data is discussed.Furthermore,the significance test of linearity and regression coefficient in transformed data is also discussed.In the end,combined with a numerical example...