Page 1 of 1
About cubic spline interpolation
Posted: Sat Dec 15, 2012 7:21 am
by magdalena
I am using Frequency Conversion copy/past from yearly series to semiannual series using Cubic-match last.
May I ask what type of natural cubic spline does EViews use.
I do compare the results with the results produced by natural cubic spline MATLAB and results are slightly different.
Re: About cubic spline interpolation
Posted: Fri Dec 28, 2012 8:25 am
by magdalena
Hi ,
Is there anybody who can help me with this? Just want to check some cubic spline results.
Re: About cubic spline interpolation
Posted: Fri Dec 28, 2012 4:04 pm
by EViews Chris
The EViews code is meant to use the conventional 'natural cubic spline' condition of no second derivative at the two end points.
However, I ran some validation checks and found that there is currently a problem in the EViews code which I expect is responsible for the differences that you're seeing.
There will be a fix for this in the next EViews patch. I'm afraid our office is shut down for the next few days, so the patch won't be available until next week.
Sorry for any confusion. Please let us know if your results still don't line up as expected after you get the updated code.
Re: About cubic spline interpolation
Posted: Fri Dec 28, 2012 11:49 pm
by magdalena
Hi Chris,
Thank you very much for your answer. I will wait for the update then, had to ask the system administrator in my office because she is the one who gets all the updates and patches.
The difference is not big but I had to know for sure why there is a difference.