[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: RE: "no observations" in kernel regression
First check that your variables are not
held by Stata as string variables. If so,
Otherwise, I presume you are using -kernreg- from STB-23,
originally published in 1996.
In 1999 I posted what I believe was a fixed
version of -kernreg- as -kernreg2-, which you can find
on SSC, given various bugs and misfeatures in -kernreg-.
Over the last five years I have encouraged
the first author of -kernreg- to publish a revised
version of -kernreg-, both in public and in private,
and both in my personal capacity and as Executive
Editor of the Stata Journal, but nothing has ever been
All this however is now history, as the program -locpoly-
(see results of -search locpoly-) essentially does
the same thing better, so long as you have Stata 8.
Otherwise try -kernreg2-, so long as you have Stata
6 or Stata 7.
Otherwise, you may need to look inside -kernreg- and see if you
can find the bug yourself.
> I am trying to run a kernel regression between variables y
> and x (independent).
> I downloaded and installed the program for the command
> kernreg from the
> Stata help file, and I am doing the following:
> kernreg yvar xvar, kercode(6) bwidth(0.65) npoint(100)
> As a result I receive the following message:
> no observations.
> I thought it had to do something with the missing values, so, to
> experiment, I kept only those observations without missing values and
> tried to run the kernel regression again. The same message appeared.
> Does anybody has a clue about why this is happening and how to solve
> this problem?
* For searches and help try: