Find the linear portion of a curve

[RandomGuy88]

Hi,

So I have a set of experimental data and I would like to find the slope of the linear portion of the data. For example, say I have set of points, the first few points are not linear, then there is a chunk of data in the middle that is linear, and then the last several points are very non-linear. I would like to find the slope of this linear portion in the middle but the location and length of the linear portion will be different for each experiment. I could easily do this by just manually removing the non-linear portions of the curve but I would like for the computer to be able to do this automatically. Does anyone have any suggestions on the best way to do this?

Thanks

 

[hotvette]

Tricky business I would think. What the human brain can easily do isn't necessarily so easy to replicate with a computer program.

http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=05716374

http://en.wikipedia.org/wiki/Robust_regression

http://www.mendeley.com/research/empirical-analysis-software-effort-estimation-outlier-elimination/

http://pubs.acs.org/doi/abs/10.1021/ed082p1472.2

 

[Bob S]

Will the curve fitting program mentioned in post #8 in https://www.physicsforums.com/showthread.php?t=364600 work?

Diffferentiate the result for the slope.

Bob S

 

[jack action]

You calculate the http://en.wikipedia.org/wiki/Differentiation_(mathematics)#Definition_via_difference_quotients" will be equal to zero (which means «no change in slope»).