I have 75 samples and want to run 500 times of linear regression analysis for it. That is, each time I can randomly pick 30 samples out of 75 and run a simple linear regression analysis, repeat 500 times but each time have different samples because it is a random pick. I want to see average values on R2, F, P, and Regression Coefficient, 95% CI.
Many thanks for you help!!!!
* The following is NOT tested, but should
* give you the general idea.
* ---- Start of macro definition ---- .
DEFINE !SimReg (NumCases = !CHAREND('/') / NumLoops = !CMDEND)
* Insert OMS commands here to direct some
* of the REGRESSION output to SPSS datasets.
!DO !i = 1 !TO !NumLoops
COMPUTE @RV@ = RV.UNIFORM(0,1).
SORT CASES by @RV@.
TEMPORARY.
SELECT IF $casenum LE !NumCases.
* Modify the following REGRESSION command as needed.
REGRESSION
/STATISTICS COEFF OUTS CI(95) R ANOVA
/DEPENDENT Y
/METHOD=ENTER X
.
!DOEND
* Insert OMSEND command here.
!ENDDEFINE.
* ---- End of macro definition ---- .
* Run the macro definition once per session
* in which it is needed.
* Call the macro.
* Delete or comment out the two SET MPRINT lines
* once you know the macro works properly.
* And test it with fewer than 500 loops!.
SET MPRINT on.
!SimReg NumCases = 75 / NumLoops = 500.
SET MPRINT off.
EDIT:
Oops! I just noticed the OP wanted to sample 30 cases each time, not 75. So the macro call should have been:
!SimReg NumCases = 30 / NumLoops = 500.
I could tell you a solution for R instead:
Given you have the data in a data.frame "df", and "model" is your model (like for instance Y~X for a simple linear regression), then this would be the code:
results = replicate(500, {
useRows = sample(1:nrow(df),size=30,replace=TRUE)
sum = summary(lm(model, data=df[useRows,] ))
c( sum$coefficients[,"Estimate"],
Rsq = sum$r.squared,
F = sum$fstatistic["value"],
p = pf(sum$fstatistic["value"],sum$fstatistic["numdf"],sum$fstatistic["dendf"]) )
})
apply(results, 1, quantile, c(0.05,0.5,0.975))
* The following is NOT tested, but should
* give you the general idea.
* ---- Start of macro definition ---- .
DEFINE !SimReg (NumCases = !CHAREND('/') / NumLoops = !CMDEND)
* Insert OMS commands here to direct some
* of the REGRESSION output to SPSS datasets.
!DO !i = 1 !TO !NumLoops
COMPUTE @RV@ = RV.UNIFORM(0,1).
SORT CASES by @RV@.
TEMPORARY.
SELECT IF $casenum LE !NumCases.
* Modify the following REGRESSION command as needed.
REGRESSION
/STATISTICS COEFF OUTS CI(95) R ANOVA
/DEPENDENT Y
/METHOD=ENTER X
.
!DOEND
* Insert OMSEND command here.
!ENDDEFINE.
* ---- End of macro definition ---- .
* Run the macro definition once per session
* in which it is needed.
* Call the macro.
* Delete or comment out the two SET MPRINT lines
* once you know the macro works properly.
* And test it with fewer than 500 loops!.
SET MPRINT on.
!SimReg NumCases = 75 / NumLoops = 500.
SET MPRINT off.
EDIT:
Oops! I just noticed the OP wanted to sample 30 cases each time, not 75. So the macro call should have been:
!SimReg NumCases = 30 / NumLoops = 500.
To me the R code looks simpler and more readable... but I might be biased :)
Jochen, I have no doubt that R (and Stata) can do this sort of thing more neatly than SPSS.
By the way, for some time, SPSS has going down the road of "programmability", which means that if one installs the needed add-ons, one can call Python or R programs from within R. I suspect that an accomplished SPSS Pythonista could write a looping Python program that is neater than a macro like the one I suggested.
@ Bin Jiang: With questions such as yours, I think you'd have better luck posting to the SPSSX-L mailing list (see link below).
http://spssx-discussion.1045642.n5.nabble.com/
All problems related to SPSS, STATA, EViews, R etc, may be solved.
Look into link by Hossian Academy below:
sayedhossain.com
Hope it is helpful
I have a problem with SPSS. I want to write a code in SPSS to repeat linear regression for ten times but every time the dependent variable change and the other part remain the same. How can do this?
I use this code to do linear regression and I want to repeat this for ten times changing the dependent variable :
REGRESSION
/MISSING LISTWISE
/STATISTICS COEFF OUTS R ANOVA
/CRITERIA=PIN(.05) POUT(.10)
/NOORIGIN
/DEPENDENT mmsecorretto
/METHOD=BACKWARD CerebralWM_FD CortexVol Mean_GM_temporal_lobe_volume sex anniscolarita
EstimatedTotalIntraCranialVol age.
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Software