Notice: On March 31, it was **announced** that Statalist is moving from an email list to a **forum**. The old list will shut down on April 23, and its replacement, **statalist.org** is already up and running.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

From |
Nick Cox <njcoxstata@gmail.com> |

To |
statalist@hsphsun2.harvard.edu |

Subject |
Re: st: Interpreting Non-Linear Least Squares |

Date |
Sat, 23 Apr 2011 17:01:00 +0100 |

You don't give the -nl- syntax you use, but 1. It is not clear that your first equation and your -nl- output line up. 2. The model you fitted is not obviously a success as -b2- has such an implausibly high t value and the R-sq is mediocre. I suspect some pathology. 3. You could reasonably post your data unless you are otherwise reluctant to do so. You should certainly give the -nl- syntax. 4. Plot the data and the fit to see what is going on! 5. You should signal whether this is some theoretical equation that evokes respect in your field or you are just interested in a good curve fit. Nick On Sat, Apr 23, 2011 at 3:58 PM, Stefan Nijssen <stefannijssen@gmail.com> wrote: > Dear Statalist users, > > In interpreting my variables, I have come upon the point to let Stata's non-linear least squares function calculate the exponent best fitting my data. However, interpreting the results I am struggling a bit. To me, the function I am looking for will be one like: > > Y = b0 + b1*(Var)^b2 > > Y being the dependent, Var the independent. Now it might very well be the case that the answer is right in front of me, however somehow to me it seems different. > > The NL least squares provides the following output, stating the estimated function has the form: > > Y = b0 + b1*b2^ebitda > > Source | SS df MS > -------------+------------------------------ Number of obs = 141 > Model | 409394.932 2 204697.466 R-squared = 0.1496 > Residual | 2327596.39 138 16866.6405 Adj R-squared = 0.1373 > -------------+------------------------------ Root MSE = 129.8716 > Total | 2736991.32 140 19549.938 Res. dev. = 1769.474 > > 3-parameter asymptotic regression, oas = b0 + b1*b2^ebitda > ------------------------------------------------------------------------------ > oas | Coef. Std. Err. t P>|t| [95% Conf. Interval] > -------------+---------------------------------------------------------------- > /b0 | -336.6996 165.5611 -2.03 0.044 -664.0642 -9.335052 > /b1 | 688.8096 178.0674 3.87 0.000 336.7163 1040.903 > /b2 | .9999999 3.13e-09 3.2e+08 0.000 .9999999 .9999999 > ------------------------------------------------------------------------------ > Parameter b0 taken as constant term in model & ANOVA table > > I feel a little stupid asking this, but can anyone give me some clues on how to read this? > * * 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/

**References**:**st: Interpreting Non-Linear Least Squares***From:*Stefan Nijssen <stefannijssen@gmail.com>

- Prev by Date:
**re: st: Interpreting Non-Linear Least Squares** - Next by Date:
**Re: st: roll rates** - Previous by thread:
**st: Interpreting Non-Linear Least Squares** - Next by thread:
**re: st: Interpreting Non-Linear Least Squares** - Index(es):