相関係数r の見方 0.7<|r|≦1 相関が強い 0.4<|r|<0.7 中間の強さ 0.2<|r|<0.4 相関が弱い 0≦|r|<0.2 相関がない
\(\normalsize\ Logarithmic\ regression\\ (1)\ mean:\ \overline{\ln x}={\large \frac{{\small \sum}{\ln x_i}}{n}},\hspace{10px}\bar{y}={\large \frac{{\small \sum}{y_i}}{n}},\hspace{10px}n={\small \sum}f_i\\ (2)\ trend\ line:\ y=A+B \ln x,\hspace{10px} B={\large\frac{Sxy}{Sxx}},\hspace{10px} A=\bar{y}-B\overline{\ln x}\\ \\ (3)\ correlation\ coefficient:\ r=\frac{\normalsize S_{xy}}{\normalsize \sqrt{S_{xx}}\sqrt{S_{yy}}}\\ \hspace{20px}S_{xx}={{\small \sum}(\ln x_i-\overline{\ln x})^2 f_i}={{\small \sum} (\ln x_i)^2 f_i} - n \cdot \overline{\ln x}^2\\ \hspace{20px}S_{yy}={{\small \sum}(y_i-\bar{y})^2 f_i}={{\small \sum} y_i^2 f_i}- n \cdot \bar{y}^2\\ \hspace{20px}S_{xy}={{\small \sum}(\ln x_i-\overline{\ln x})(y_i-\bar{y}) f_i}={{\small \sum} \ln x_i y_i f_i}- n \cdot \overline{\ln x}\bar{y}\\ \) |
|