Error fitting of a slope in mathcad

[FrankButcher]

Hello,

I have a submission tomorrow and I need to calculate the error in my calculated value of the intercept c and gradient m.

I have used both the mathcad function and the manual version of it to fit my data points to a line (finding alpha and beta for a langmuir isotherm) and now just need to use the error values for my measurements to get the final absolute error in the slope and the intercept. Can anybody help?

The data set sigma corresponds to the error in eta, the excess amount adsorbed.

Cheers

Fraser

[Werner_E]

Use stderr(P,P.n)=

Via help you can access a helpul quicksheet about Linear Regression.

[FrankButcher]

Thanks for the quick response Werner,

I have already tried this however I want to know the error attributed to the propagation of the uncertainties in my values of eta.

Do you have any idea how to do this mathetmatically?

Also I have calculated errors for another part of the paper, however they are errors in a value of x. I have plotted x against a varying y and don't seem to be able to add error bars that represent the errors in the value of x; how would I go about doing this? See attached.

Cheers

Fraser

[Werner_E]

I have already tried this however I want to know the error attributed to the propagation of the uncertainties in my values of eta.

I am not sure what exactly this error should measure and how you would define it.

[FrankButcher]

The values used to calculate the intercept and slope of the line are gathered from a simulation and as such each data point has an individual uncertainty attributed to it. I would like to correlate these uncertainties to get the maximum error in the calculated values of y-intercept and gradient.

Is this possible analytically?

[Werner_E]

As I understand you just want to know how you values of eta influence the aberration of (P; Pn) from a perfect straight line!?

You get a straight line if you vector eta consists of numers all equal.

You you have to set up a measure for the aberration of a vector from a constant vector.

Maybe something like this???