Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
| From | "Nick Cox" <n.j.cox@durham.ac.uk> |
| To | <statalist@hsphsun2.harvard.edu> |
| Subject | st: RE: RE: RE: missing values in panel data |
| Date | Wed, 4 Aug 2010 16:09:08 +0100 |
I'd try interpolation on logarithmic scale subject to all values being positive. Nick n.j.cox@durham.ac.uk Schöler, Lisa I have tried -ipolate- now. When I look at the correlation between the predicted variables and the "real" variables for the years where I have the data the correlation is only .25 which is not very satisfying. Any other suggestion what could make sense? Nick Cox I don't know what is _best_, but I don't see why you think using -ipolate- is not possible, as -ipolate- can also be used for extrapolation. (I don't think there is any command -interpolate-.) Schöler, Lisa in my panel data set (unbalanced) I have a variable called "RD spendings". I have those data for the years 2001-2009 (sometimes less years) for 200 companies each year. Now I need to predict/forecast the missing data for 1999 and 2000. I think that -interpolate- is not possible. Looking at the trend between 2001 and 2009, I can tell that the spendings are rising most of the times. Can somebody give me a suggestion what would be the best solution with Stata? * * For searches and help try: * http://www.stata.com/help.cgi?search * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/