Deriving the Least-Squares Estimates for Simple Linear Regression
The following supplemental notes were created by Dr. Maria Tackett for STA 210. They are provided for students who want to dive deeper into the mathematics behind regression and reflect some of the material covered in STA 211: Mathematics of Regression. Additional supplemental notes will be added throughout the semester.
This document contains the mathematical details for deriving the least-squares estimates for slope (\(\beta_1\)) and intercept (\(\beta_0\)). We obtain the estimates, \(\hat{\beta}_1\) and \(\hat{\beta}_0\) by finding the values that minimize the sum of squared residuals, as shown in Equation 1.
\[ SSR = \sum\limits_{i=1}^{n}[y_i - \hat{y}_i]^2 = [y_i - (\hat{\beta}_0 + \hat{\beta}_1 x_i)]^2 = [y_i - \hat{\beta}_0 - \hat{\beta}_1 x_i]^2 \qquad(1)\]
Recall that we can find the values of \(\hat{\beta}_1\) and \(\hat{\beta}_0\) that minimize /eq-ssr by taking the partial derivatives of Equation 1 and setting them to 0. Thus, the values of \(\hat{\beta}_1\) and \(\hat{\beta}_0\) that minimize the respective partial derivative also minimize the sum of squared residuals. The partial derivatives are shown in Equation 2.
\[ \begin{aligned} \frac{\partial \text{SSR}}{\partial \hat{\beta}_1} &= -2 \sum\limits_{i=1}^{n}x_i(y_i - \hat{\beta}_0 - \hat{\beta}_1 x_i) \\ \frac{\partial \text{SSR}}{\partial \hat{\beta}_0} &= -2 \sum\limits_{i=1}^{n}(y_i - \hat{\beta}_0 - \hat{\beta}_1 x_i) \end{aligned} \qquad(2)\]
The derivation of deriving \(\hat{\beta}_0\) is shown in Equation 3.
\[ \begin{aligned}\frac{\partial \text{SSR}}{\partial \hat{\beta}_0} &= -2 \sum\limits_{i=1}^{n}(y_i - \hat{\beta}_0 - \hat{\beta}_1 x_i) = 0 \\&\Rightarrow -\sum\limits_{i=1}^{n}(y_i + \hat{\beta}_0 + \hat{\beta}_1 x_i) = 0 \\&\Rightarrow - \sum\limits_{i=1}^{n}y_i + n\hat{\beta}_0 + \hat{\beta}_1\sum\limits_{i=1}^{n}x_i = 0 \\&\Rightarrow n\hat{\beta}_0 = \sum\limits_{i=1}^{n}y_i - \hat{\beta}_1\sum\limits_{i=1}^{n}x_i \\&\Rightarrow \hat{\beta}_0 = \frac{1}{n}\Big(\sum\limits_{i=1}^{n}y_i - \hat{\beta}_1\sum\limits_{i=1}^{n}x_i\Big)\\&\Rightarrow \hat{\beta}_0 = \bar{y} - \hat{\beta}_1 \bar{x} \\\end{aligned} \qquad(3)\]
The derivation of \(\hat{\beta}_1\) using the \(\hat{\beta}_0\) we just derived is shown in Equation 4.
\[ \begin{aligned}&\frac{\partial \text{SSR}}{\partial \hat{\beta}_1} = -2 \sum\limits_{i=1}^{n}x_i(y_i - \hat{\beta}_0 - \hat{\beta}_1 x_i) = 0 \\&\Rightarrow -\sum\limits_{i=1}^{n}x_iy_i + \hat{\beta}_0\sum\limits_{i=1}^{n}x_i + \hat{\beta}_1\sum\limits_{i=1}^{n}x_i^2 = 0 \\\text{(Fill in }\hat{\beta}_0\text{)}&\Rightarrow -\sum\limits_{i=1}^{n}x_iy_i + (\bar{y} - \hat{\beta}_1\bar{x})\sum\limits_{i=1}^{n}x_i + \hat{\beta}_1\sum\limits_{i=1}^{n}x_i^2 = 0 \\&\Rightarrow (\bar{y} - \hat{\beta}_1\bar{x})\sum\limits_{i=1}^{n}x_i + \hat{\beta}_1\sum\limits_{i=1}^{n}x_i^2 = \sum\limits_{i=1}^{n}x_iy_i \\&\Rightarrow \bar{y}\sum\limits_{i=1}^{n}x_i - \hat{\beta}_1\bar{x}\sum\limits_{i=1}^{n}x_i + \hat{\beta}_1\sum\limits_{i=1}^{n}x_i^2 = \sum\limits_{i=1}^{n}x_iy_i \\&\Rightarrow n\bar{y}\bar{x} - \hat{\beta}_1n\bar{x}^2 + \hat{\beta}_1\sum\limits_{i=1}^{n}x_i^2 = \sum\limits_{i=1}^{n}x_iy_i \\&\Rightarrow \hat{\beta}_1\sum\limits_{i=1}^{n}x_i^2 - \hat{\beta}_1n\bar{x}^2 = \sum\limits_{i=1}^{n}x_iy_i - n\bar{y}\bar{x} \\&\Rightarrow \hat{\beta}_1\Big(\sum\limits_{i=1}^{n}x_i^2 -n\bar{x}^2\Big) = \sum\limits_{i=1}^{n}x_iy_i - n\bar{y}\bar{x} \\ &\hat{\beta}_1 = \frac{\sum\limits_{i=1}^{n}x_iy_i - n\bar{y}\bar{x}}{\sum\limits_{i=1}^{n}x_i^2 -n\bar{x}^2}\end{aligned} \qquad(4)\]
To write \(\hat{\beta}_1\) in a form that’s more recognizable, we will use the following:
\[ \sum x_iy_i - n\bar{y}\bar{x} = \sum(x - \bar{x})(y - \bar{y}) = (n-1)\text{Cov}(x,y) \qquad(5)\]
\[ \sum x_i^2 - n\bar{x}^2 - \sum(x - \bar{x})^2 = (n-1)s_x^2 \qquad(6)\]
where \(\text{Cov}(x,y)\) is the covariance of \(x\) and \(y\), and \(s_x^2\) is the sample variance of \(x\) (\(s_x\) is the sample standard deviation).
Thus, applying Equation 5 and Equation 6, we have
\[ \begin{aligned}\hat{\beta}_1 &= \frac{\sum\limits_{i=1}^{n}x_iy_i - n\bar{y}\bar{x}}{\sum\limits_{i=1}^{n}x_i^2 -n\bar{x}^2} \\&= \frac{\sum\limits_{i=1}^{n}(x-\bar{x})(y-\bar{y})}{\sum\limits_{i=1}^{n}(x-\bar{x})^2}\\&= \frac{(n-1)\text{Cov}(x,y)}{(n-1)s_x^2}\\&= \frac{\text{Cov}(x,y)}{s_x^2}\end{aligned} \qquad(7)\]
The correlation between \(x\) and \(y\) is \(r = \frac{\text{Cov}(x,y)}{s_x s_y}\). Thus, \(\text{Cov}(x,y) = r s_xs_y\). Plugging this into Equation 7, we have
\[ \hat{\beta}_1 = \frac{\text{Cov}(x,y)}{s_x^2} = r\frac{s_ys_x}{s_x^2} = r\frac{s_y}{s_x} \qquad(8)\]