Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

RE: st: RE: Replacing missing values


From   "Nick Cox" <n.j.cox@durham.ac.uk>
To   <statalist@hsphsun2.harvard.edu>
Subject   RE: st: RE: Replacing missing values
Date   Thu, 22 Apr 2004 12:26:58 +0100

I can't say what would be better. Tritely, 
interpolation works well when you have a few 
gaps in very smooth series, and no crude 
interpolation method works well otherwise.  

I wrote -cipolate- on SSC partly for 
a climatological problem (gaps in 
temperature time series) for which -ipolate- 
seemed too crude, but in practice my tests 
based on omitting real data randomly and 
seeing which gave the better reconstruction 
did not in fact indicate much better performance
by -cipolate-. That is not very surprising
mathematically for small gaps. 

I'll stress (again) for anyone listening 
that -cipolate- is not a spline method. 
That is, cubic interpolation is not 
cubic spline interpolation. 

You have to try it and see. No remote 
counselling can say what is better for 
your data. 

Nick 
n.j.cox@durham.ac.uk 

joe J.
 
> Thanks Nick. True, linear interpolation is not a great idea. 
> Another option 
> is to regress investment series on related variables (eg. 
> fuel) and time 
> trend and use the fitted  values to replace missing values. 
> But I think this 
> would be okay if I am dealing with capital stock rather than 
> investment.
> 
> In fact I toyed with the idea of cubic interpolation, and 
> wonder if the 
> fluctuating character of investment decisions is a 
> justification for using 
> it?

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index