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

RE: st: RE: Brute force

From   "Verkuilen, Jay" <>
To   <>
Subject   RE: st: RE: Brute force
Date   Mon, 16 Mar 2009 18:53:10 -0400

I agree, I think the term "grid search" is much more applicable. Many nonlinear regression programs allow grid search as an option (and rightly so).

Using initial values, iteration limits and some for loops, it should be possible to do this in Stata without too much trouble.


-----Original Message-----
From: [] On Behalf Of Maarten buis
Sent: Monday, March 16, 2009 4:53 PM
Subject: Re: st: RE: Brute force

--- On Mon, 16/3/09, David Airey wrote:
> This might be what he is asking...?

looks like a grid search to me, which would be a 
special case of a "brute force" method. This method
could be useful to assess the sensitivity of your
results to different staring values, which makes
sense with a method like nonlinear least squares.
However, I would only apply it after a regular 
nonlinear least squares estimation (-nl- in Stata).

A different question is whether this method would
deserve its own name. Assigning names to simple
techniques often just add more confusion than 
clarity. In this case you could just write in a
paper "I tried out different starting values and
the results did (not) change". If you insisted 
on calling it the "brute force algorithm", than 
the resuting text would sound more impressive 
but would actually be less clear: "I applied 
the brute force algorithm and the results proved
(not) to be robust"

-- Maarten

Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen


*   For searches and help try:

*   For searches and help try:

© Copyright 1996–2021 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index