# st: RE: Curve Fit

 From "FEIVESON, ALAN H. (AL) (JSC-SD) (NASA)" <[email protected]> To "'[email protected]'" <[email protected]> Subject st: RE: Curve Fit Date Mon, 9 Dec 2002 12:50:44 -0600

```Marita -

I'm not totally sure what you are asking - but it sounds as if you want to
regress your dependent variable on a pre-defined function. To be specific,
log(x), where x is also a Stata variable. Then you could do the following:

.gen fx=log(x)
.reg y fx

This will find a linear transformation of log x (not x) that best fits your
y-variable in a least-squares sense. In other words, y is predicted by b0
+b1*log(x). Note that this model is linear in the coefficients (b0, b1) even
though log(x) is a nonlinear function of x.

On the other hand if you want to predict y by a model that is nonlinear in
the coefficients, such as b0*exp(-b1*x)+b2, then you have to use nonlinear
least squares (the -nl- command).

Hope this helps -

Al Feiveson

-----Original Message-----
From: Marita Jacob [mailto:[email protected]]
Sent: Monday, December 09, 2002 7:50 AM
To: [email protected]
Subject: st: Curve Fit

Hi Stata-Users,

is it possible to use Stata to fit a selected self-defined curve to a line
plot?
I would like to define my own model y=f(x) to be fit to the data (which is
not necessarily the best fitting function, but I am interested in the
particular form of the curve).

Thanks,
Marita Jacob

*
*   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/
*
*   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/
```