Subroutine to run several linear regressions and to compute mean of coefficients

Subroutine to run several linear regressions and to compute mean of coefficients

Post by Ernie » Sat, 19 Jul 2003 03:37:09



hi all,

in order to validate the coefficients I have found for my sample of
175 items I want to run a routine that

a) draws subsamples of 50 items, computes the coeffcients for a linear
regression with 12 x's
b) saves the coefficients, i.e. adds them to the previously found
coefficients
c) quits after , say, 20 runs,
and finally computes the mean for each coefficient.

the idea behind this is that if the mean is more or less equal to
coefficient for the whole sample, the coefficient can be assumed to be
"correct" in general.

does anyone have any clue how to do this in SPSS ?

thx

 
 
 

Subroutine to run several linear regressions and to compute mean of coefficients

Post by Rich Ulric » Sat, 19 Jul 2003 04:45:38



Quote:> hi all,

> in order to validate the coefficients I have found for my sample of
> 175 items I want to run a routine that

> a) draws subsamples of 50 items, computes the coeffcients for a linear
> regression with 12 x's
> b) saves the coefficients, i.e. adds them to the previously found
> coefficients
> c) quits after , say, 20 runs,
> and finally computes the mean for each coefficient.

[ ... ]

This is called "bootstrapping"  and you can find some
examples when you search under that word.
Using  group.google -

  bootstrap OR bootstrapping   group:comp.soft-sys.stat.spss

--

http://www.pitt.edu/~wpilib/index.html
"Taxes are the price we pay for civilization."  Justice Holmes.

 
 
 

1. EXCEL5.0 regression analysis: How to get regression coefficients?

Hi!

Using VBA, one can easily tell Excel to plot a trend line using  
exponential/linear/polynomial regression and it is easy to tell the  
program to include the "best fit" coefficients of the used regression  
formula in the chart.
However, I couldn't find a way to fetch those coefficients in VBA into a  
variable (e.g.: to apply some different sorts of regression formulae and  
decide which one to use for further computations).
--Any clues?

--


Feed the BND-cookie reader!: Mord, Terror, Drogen, Steuer, Geld, Betrug
Demokratie, Freiheit, Fernmeldegeheimnis, Kryptographie, Liberalismus

2. Terminology clarification re: DECwindows et.al.

3. Non-homogeneous system of linear 2nd-order differential equations with constant coefficients

4. Deleted UserName-Need to re-activate

5. coefficient matrix from a set of linear equations

6. What's the fastest 4 mm Dat Drive ?

7. linear-by-linear association in log-linear model

8. Distance finder

9. Comparing regression coefficients to a known value

10. standardised coefficients for logistic regression

11. Rounding regression coefficients

12. outputting regression coefficients in data set?

13. Comparing regression coefficients to a known value