| T O P I C R E V I E W |
| CCKammerer |
Posted - 06/10/2012 : 08:41:09 AM
Origin Ver. and Service Release (Select Help-->About Origin): 8.6SR3
Operating System: Windows7
Is there a way to do a cubic spline fit where I can graphically manipulate the position of the nodes or junctions? Also, I would like to see the fit statistics.
C.C.Kammerer |
| 2 L A T E S T R E P L I E S (Newest First) |
| Sam Fang |
Posted - 06/21/2012 : 03:19:01 AM
NAG function e02bac supports to change the interior knots of B spline, i.e. spline.lamda. You can use it to graphically manipulate the position of spline knots.
Sam
OriginLab Technical Services |
| Kathy_Wang |
Posted - 06/11/2012 : 06:09:25 AM
Hi,
In Origin, you may call a NAG function (e02bac) to do a cubic spline fit with nodes, and also get the residual sum square as the fit statistics.
For details of this NAG function, you may refer to the following PDF:
http://www.originlab.com/pdfs/nagcl07/manual/pdf/e02/e02bac.pdf
And there is a tutorial showing how to call NAG function in Origin C.
http://wiki.originlab.com/~originla/howto/index.php?title=Tutorial:Calling_NAG_Functions_From_Origin_C
Also, the built-in method "Cubic B-spline"(using e02bec) for interpolation might also fulfill your needs. But this function does the fitting in a slightly different way, that the fitting is to minimize the discontinuity jump in the third order derivative rather than minimize the residual when calling the e02bac function.
You could find the details of e02bec in the following link
http://www.originlab.com/pdfs/nagcl07/manual/pdf/e02/e02bec.pdf
In addition, when you use the method below, you move control points instead of spline knots to change the fitted curve.
To use this build-in function, you could first plot your data, and select from top menu "Analysis:Mathematics:Interpolates/Extrapolates" to open the interpolation dialog, and do the setting as following figure:
Note that the smoothing factor is controlling the smoothness of your fitted curve. The bigger this factor is, the smoother your fitted curve will be, but also relevantly the bigger the residual will be.
After the fitting, you will get the result figure like this:
Each symbol is one nodes/junctions, and you may select one of the data point by clicking on it twice(but not double click), and select from top menu "Data:Move Data Points" and move the data point to somewhere else, and then click the recalculate button (or you could set recalculate mode to Auto to make it update automatically after data points change)
Hope this information helps!
Kathy
|
|
|
