Standard error of regression

#1
I am writing a robust regression algorith in Python and using the linregress method from SciPy/NumPy. linregress returns the b0, b1, r^2, p_val, and standard error of the estimate.

I need to devise the standard error of the regression without looping through to add up the error terms. Is there a calculation that "translations" one or all of these results into the standard error of the regression?
 

Dragan

Super Moderator
#2
I am writing a robust regression algorith in Python and using the linregress method from SciPy/NumPy. linregress returns the b0, b1, r^2, p_val, and standard error of the estimate.

I need to devise the standard error of the regression without looping through to add up the error terms. Is there a calculation that "translations" one or all of these results into the standard error of the regression?
For a simple regression (which is what it looks like you have) it is:

Stderror = Sy* Sqrt[ (1 - R^2)* [(N - 1)/(N - 2)] ]

where Sy is the standard deviation of Y.