- #1
mikelee8a
- 2
- 0
Hi,
I'd like to fit a straight line to some data which is noisey with gaussian noise with some st dev.
Using least squares, I can estimate the slope and intercept. I'd like to know the uncertainty in these numbers. I can find the residue, I believe this is a measure of the variance of the noise.
Using a simulation with N points, each with noise st dev. σ, I find the variation in estimated slope is proportional to σ/(√N^3), which I can't explain, I'd have expected sigma over root N, as for the standard error.
Any help would be fantastic. I just want to know what error to quote with my fitted gradient.
Mike
I'd like to fit a straight line to some data which is noisey with gaussian noise with some st dev.
Using least squares, I can estimate the slope and intercept. I'd like to know the uncertainty in these numbers. I can find the residue, I believe this is a measure of the variance of the noise.
Using a simulation with N points, each with noise st dev. σ, I find the variation in estimated slope is proportional to σ/(√N^3), which I can't explain, I'd have expected sigma over root N, as for the standard error.
Any help would be fantastic. I just want to know what error to quote with my fitted gradient.
Mike