How do I get RSS from a linear model output

wszsdmjj picture wszsdmjj · Nov 7, 2016 · Viewed 38.7k times · Source

Below is a linear model output for a dataset consisting of a response variable and three explanatory variables. How do I get the RSS of the original regression?

  Call:
  lm(formula = y ~ x1 + x2 + x3)
Residuals:
      Min      1Q  Median      3Q     Max
  -4.9282 -1.3174  0.0059  1.3238  4.4560
  Coefficients:
               Estimate Std. Error t value Pr(>|t|)
  (Intercept) -7.056057   1.963805  -3.593 0.000481 ***
  x1           3.058592   0.089442  34.196  < 2e-16 ***
  x2          -5.763410   0.168072 -34.291  < 2e-16 ***
  x3           0.000571   0.165153   0.003 0.997247
  ---
  Signif. codes:  0 *** 0.001 ** 0.01 * 0.05 . 0.1   1
  Residual standard error: 1.928 on 116 degrees of freedom

Multiple R-squared:  0.9546,Adjusted R-squared:  0.9535
F-statistic:   814 on 3 and 116 DF,  p-value: < 2.2e-16

Answer

G. Grothendieck picture G. Grothendieck · Nov 7, 2016

Here are some ways of computing the residual sum of squares (RSS) using the built-in anscombe data set:

fm <- lm(y1 ~ x1+x2+x3, anscombe)

deviance(fm)
## [1] 13.76269

sum(resid(fm)^2)
## [1] 13.76269

anova(fm) # see the Residuals row of the Sum Sq column
## Analysis of Variance Table
##
## Response: y1
##           Df Sum Sq Mean Sq F value  Pr(>F)   
## x1         1 27.510 27.5100   17.99 0.00217 **
## Residuals  9 13.763  1.5292                   
## ---
## Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

anova(fm)["Residuals", "Sum Sq"]
## [1] 13.76269

with(summary(fm), df[2] * sigma^2)
## [1] 13.76269

Regarding the last one, note that summary(fm)$df[2] and summary(fm)$sigma are shown in the summary(fm) output in case you want to calculate RSS using only a printout from summary. In particular, for the output shown in the question df[2] = 116 and sigma = 1.928 so RSS = df[2] * sigma^2 = 116 * 1.928^2 = 431.1933 .