r - The standard deviation of the absolute error for a linear regression model -

in linear regression model, have observed_values , predicted_values. want calculate standard deviation of absolute error values in r. think this, not sure:

sd(abs(observed_values-predicted_values)) 

is o.k.? there sort of function that?

suppose linear model fit lmfit, need do:

n <- length(lmfit$residuals)  ## number of data / residuals df.residual <- lmfit$df.residual  ## residual degree of freedom abs.residual <- abs(lmfit$residuals)  ## absolute residuals 

now, sample standard deviation sd(abs.residual) biased estimate, because assumes n-1 degree of freedom in residuals. while in fact, there df.residual degree of freedom. need bias correction:

sd(abs.residual) * sqrt((n-1) / df.residual)