Origin Version (Select Help-->About Origin): Operating System: Windows XP The Analysis\Calculate\differentiate function can do differentiation. Unfortunately, if the neighboring data points contain error, the obtained derivative will enlarge the error, and contains too much noise. Sometimes the derivative may be too large (infinite, for example) or too small (0, for example) and its sign may be wrong. However, the general trend of my curve is very smooth and its derivative should be well-defined and definite. By trial-and-error method, I found the following method to solve the problem Use Analysis\Interpolate/Extrapolate to generate another set of data points that reflects the general smooth trend of the original curve; Differentiate the curve plotted using the 2nd set of data.
My questions are (1) How to ensure the 2nd data set "resemble" the original data set? Specifically, how many data points are the best in the 2nd data set? (2) Is there any other, better method? |