SPSSX Discussion

saving MATRIX output to SAV file

Classic

List

Threaded

11 messages Options

Beckstead, Jason

saving MATRIX output to SAV file

I am using MATRIX/ END MATRIX to calculate the determinant of a correlation matrix. In my data file I have 6 variables, each is a correlation coefficient from a sample. Each row of the data file is from a different sample created as part of a Monte Carlo simulation, so the file has about 10,000 rows. I need to select a row (6 correlations for a sample) pass the values to MATRIX get the determinant, and then write the sample id and the value of the determinant to an external SAV file that I can then match to the original data file.

I put the MATRIX code into a macro loop which runs, but the SAVE command overwrites rather than appends the results.

My code so far looks like this.

GET FILE='CNTR_SIM3_12CONDS.SAV'

/KEEP= rYX1u rYX2u rYpdctu rX1X2u rX1pdctu rX2pdctu SAMPLE_.

SELECT IF SAMPLE_<=100.

DEFINE !DETS (!POS !TOKENS(1)).

!DO !I = 1 !TO !1.

TEMP.

SELECT IF SAMPLE_=!I.

MATRIX.

GET VALS.

COMPUTE R=MAKE(4,4,1).

COMPUTE R(2,1)=VALS(1).

COMPUTE R(3,1)=VALS(2).

COMPUTE R(4,1)=VALS(3).

COMPUTE R(3,2)=VALS(4).

COMPUTE R(4,2)=VALS(5).

COMPUTE R(4,3)=VALS(6).

COMPUTE R(1,2)=VALS(1).

COMPUTE R(1,3)=VALS(2).

COMPUTE R(1,4)=VALS(3).

COMPUTE R(2,3)=VALS(4).

COMPUTE R(2,4)=VALS(5).

COMPUTE R(3,4)=VALS(6).

COMPUTE SAMPLE_=VALS(7).

COMPUTE D=DET(R).

COMPUTE RST={SAMPLE_,D}.

PRINT RST.

SAVE RST/OUTFILE='DETS.SAV'

/VARIABLES=SMP DET.

END MATRIX.

!DOEND.

!ENDDEFINE.

!DETS 100.

Any help would be much appreciated.

Jason

_____________________________________________________________

Jason W. Beckstead, Ph.D.

Associate Professor/Quantitative Methodologist

University of South Florida College of Nursing

12901 Bruce B. Downs Blvd., MDC22, Tampa, FL 33612, USA

Statistical Editor, International Journal of Nursing Studies

phone: +1.813.974.7667 fax: +1.813.974.5418

personal website: http://personal.health.usf.edu/jbeckste/

International Journal of Nursing Studies http://www.elsevier.com/ijns

===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD

Kirill Orlov

Re: saving MATRIX output to SAV file

Of course. You are looping whole MATRIX sessions, that will not do.
Instead, you may put just SAVE statement inside !DO - !DOEND or (better) inside regular LOOP - END LOOP.
New cases will be appended in loops to an output sav file.

21.02.2015 19:52, Beckstead, Jason пишет:

I am using MATRIX/ END MATRIX to calculate the determinant of a correlation matrix. In my data file I have 6 variables, each is a correlation coefficient from a sample. Each row of the data file is from a different sample created as part of a Monte Carlo simulation, so the file has about 10,000 rows. I need to select a row (6 correlations for a sample) pass the values to MATRIX get the determinant, and then write the sample id and the value of the determinant to an external SAV file that I can then match to the original data file.
I put the MATRIX code into a macro loop which runs, but the SAVE command overwrites rather than appends the results.

My code so far looks like this.

GET FILE='CNTR_SIM3_12CONDS.SAV'

/KEEP= rYX1u rYX2u rYpdctu rX1X2u rX1pdctu rX2pdctu SAMPLE_.

SELECT IF SAMPLE_<=100.

DEFINE !DETS (!POS !TOKENS(1)).

!DO !I = 1 !TO !1.

TEMP.

SELECT IF SAMPLE_=!I.

MATRIX.

GET VALS.

COMPUTE R=MAKE(4,4,1).

COMPUTE R(2,1)=VALS(1).

COMPUTE R(3,1)=VALS(2).

COMPUTE R(4,1)=VALS(3).

COMPUTE R(3,2)=VALS(4).

COMPUTE R(4,2)=VALS(5).

COMPUTE R(4,3)=VALS(6).

COMPUTE R(1,2)=VALS(1).

COMPUTE R(1,3)=VALS(2).

COMPUTE R(1,4)=VALS(3).

COMPUTE R(2,3)=VALS(4).

COMPUTE R(2,4)=VALS(5).

COMPUTE R(3,4)=VALS(6).

COMPUTE SAMPLE_=VALS(7).

COMPUTE D=DET(R).

COMPUTE RST={SAMPLE_,D}.

PRINT RST.

SAVE RST/OUTFILE='DETS.SAV'

/VARIABLES=SMP DET.

END MATRIX.

!DOEND.

!ENDDEFINE.

!DETS 100.

Any help would be much appreciated.

Jason

_____________________________________________________________

Jason W. Beckstead, Ph.D.

Associate Professor/Quantitative Methodologist

University of South Florida College of Nursing

12901 Bruce B. Downs Blvd., MDC22, Tampa, FL 33612, USA

Statistical Editor, International Journal of Nursing Studies

phone: +1.813.974.7667 fax: +1.813.974.5418

personal website:  http://personal.health.usf.edu/jbeckste/

International Journal of Nursing Studies  http://www.elsevier.com/ijns

===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD

David Marso

Re: saving MATRIX output to SAV file

Administrator

Assuming the cases are ordered with sample 1 being rows 1-6, sample 2 rows 7-12 etc.
Untested...
may need to fiddle a bit to defray golden monkey residuals... ;-)

MATRIX.
GET data /FILE * /VARIABLES sample corr.
LOOP I=1 TO TRUNC(NROW(data)/6).
COMPUTE R=MAKE(4,4,0).
COMPUTE start=(I-1)*6+1.
COMPUTE R(1,2)=data(start).
COMPUTE R(1,3)=data(start+1).
.....
COMPUTE R(3,4)=data(start+6).
COMPUTE R=R+T(R)+ MDIAG(4,1).
SAVE {I,DET(R)}/OUTFILE "filepath" /VARIABLES sample Det .
END LOOP.
END MATRIX.

Kirill Orlov wrote

Of course. You are looping whole MATRIX sessions, that will not do.
Instead, you may put just SAVE statement inside !DO - !DOEND or (better)
inside regular LOOP - END LOOP.
New cases will be appended in loops to an output sav file.

21.02.2015 19:52, Beckstead, Jason пишет:
> I am using MATRIX/ END MATRIX to calculate the determinant of a
> correlation matrix. In my data file I have 6 variables, each is a
> correlation coefficient from a sample. Each row of the data file is
> from a different sample created as part of a Monte Carlo simulation,
> so the file has about 10,000 rows. I need to select a row (6
> correlations for a sample) pass the values to MATRIX get the
> determinant, and then write the sample id and the value of the
> determinant to an external SAV file that I can then match to the
> original data file.
> I put the MATRIX code into a macro loop which runs, but the SAVE
> command overwrites rather than appends the results.
> My code so far looks like this.
>
> GET FILE='CNTR_SIM3_12CONDS.SAV'
> /KEEP= rYX1u rYX2u rYpdctu rX1X2u rX1pdctu rX2pdctu SAMPLE_.
> SELECT IF SAMPLE_<=100.
>
> DEFINE !DETS (!POS !TOKENS(1)).
> !DO !I = 1 !TO !1.
> TEMP.
> SELECT IF SAMPLE_=!I.
> MATRIX.
> GET VALS.
> COMPUTE R=MAKE(4,4,1).
> COMPUTE R(2,1)=VALS(1).
> COMPUTE R(3,1)=VALS(2).
> COMPUTE R(4,1)=VALS(3).
> COMPUTE R(3,2)=VALS(4).
> COMPUTE R(4,2)=VALS(5).
> COMPUTE R(4,3)=VALS(6).
> COMPUTE R(1,2)=VALS(1).
> COMPUTE R(1,3)=VALS(2).
> COMPUTE R(1,4)=VALS(3).
> COMPUTE R(2,3)=VALS(4).
> COMPUTE R(2,4)=VALS(5).
> COMPUTE R(3,4)=VALS(6).
> COMPUTE SAMPLE_=VALS(7).
> COMPUTE D=DET(R).
> COMPUTE RST={SAMPLE_,D}.
> PRINT RST.
> SAVE RST/OUTFILE='DETS.SAV'
> /VARIABLES=SMP DET.
> END MATRIX.
> !DOEND.
> !ENDDEFINE.
> !DETS 100.
>
> Any help would be much appreciated.
>
> Jason
> _____________________________________________________________
>
> Jason W. Beckstead, Ph.D.
>
> Associate Professor/Quantitative Methodologist
>
> University of South Florida College of Nursing
>
> 12901 Bruce B. Downs Blvd., MDC22, Tampa, FL 33612, USA
>
> Statistical Editor, International Journal of Nursing Studies
>
> phone: +1.813.974.7667 fax: +1.813.974.5418
>
> personal website: http://personal.health.usf.edu/jbeckste/
>
> International Journal of Nursing Studies http://www.elsevier.com/ijns
>
>
> ===================== To manage your subscription to SPSSX-L, send a
> message to [hidden email]
> <mailto:[hidden email]> (not to SPSSX-L), with no body text
> except the command. To leave the list, send the command SIGNOFF
> SPSSX-L For a list of commands to manage subscriptions, send the
> command INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

Please reply to the list and not to my personal email.
Those desiring my consulting or training services please feel free to email me.
---
"Nolite dare sanctum canibus neque mittatis margaritas vestras ante porcos ne forte conculcent eas pedibus suis."
Cum es damnatorum possederunt porcos iens ut salire off sanguinum cliff in abyssum?"

Beckstead, Jason

Re: saving MATRIX output to SAV file

In reply to this post by Beckstead, Jason

David and others,

thanks for your suggestions. I tried adapting your code but I must be missing something with the LOOP in MATRIX procedure. I keep getting an undefined error. My file structure is wide, rather than tall so all 6 vars are on the same line for each sample, but I don't think that is the problem. I think it is with the LOOP because I can't even get the code to print 3 R mtxs. Here is the set up for the first 3 of 10,000 samples.

DATA LIST FREE/rYX1 rYX2 rYpdct rX1X2 rX1pdct rX2pdct SAMPLE_.

BEGIN DATA.

.59 .41 .71 -.05 .67 .70 1.00

.49 .35 .62 -.06 .79 .55 2.00

.55 .38 .67 -.06 .78 .56 3.00

END DATA.

MATRIX.

GET data/FILE=*.

LOOP i=1 TO NROW(data).

COMPUTE R=MAKE(4,4,1).

COMPUTE R(2,1)=data(1).

COMPUTE R(3,1)=data(2).

COMPUTE R(4,1)=data(3).

COMPUTE R(3,2)=data(4).

COMPUTE R(4,2)=data(5).

COMPUTE R(4,3)=data(6).

COMPUTE R(1,2)=data(1).

COMPUTE R(1,3)=data(2).

COMPUTE R(1,4)=data(3).

COMPUTE R(2,3)=data(4).

COMPUTE R(2,4)=data(5).

COMPUTE R(3,4)=data(6).

PRINT R.

SAVE {I,DET(R)} /OUTFILE='DETS.SAV' /VARIABLES=SAMPLE_ DET.

END LOOP.

END MATRIX.

Any further insight would be appreciated.

Jason

_____________________________________________________________

Jason W. Beckstead, Ph.D.

Associate Professor/Quantitative Methodologist

University of South Florida College of Nursing

12901 Bruce B. Downs Blvd., MDC22, Tampa, FL 33612, USA

Statistical Editor, International Journal of Nursing Studies

phone: +1.813.974.7667 fax: +1.813.974.5418

personal website: http://personal.health.usf.edu/jbeckste/

International Journal of Nursing Studies http://www.elsevier.com/ijns

David Marso

Re: saving MATRIX output to SAV file

Administrator

You want data(I,1) for the first assignment and I in the others as well.

Beckstead, Jason wrote

David and others,
thanks for your suggestions. I tried adapting your code but I must be missing something with the LOOP in MATRIX procedure. I keep getting an undefined error. My file structure is wide, rather than tall so all 6 vars are on the same line for each sample, but I don't think that is the problem. I think it is with the LOOP because I can't even get the code to print 3 R mtxs. Here is the set up for the first 3 of 10,000 samples.

DATA LIST FREE/rYX1 rYX2 rYpdct rX1X2 rX1pdct rX2pdct SAMPLE_.
BEGIN DATA.
.59 .41 .71 -.05 .67 .70 1.00
.49 .35 .62 -.06 .79 .55 2.00
.55 .38 .67 -.06 .78 .56 3.00
END DATA.
MATRIX.
GET data/FILE=*.
LOOP i=1 TO NROW(data).
COMPUTE R=MAKE(4,4,1).
COMPUTE R(2,1)=data(1).
COMPUTE R(3,1)=data(2).
COMPUTE R(4,1)=data(3).
COMPUTE R(3,2)=data(4).
COMPUTE R(4,2)=data(5).
COMPUTE R(4,3)=data(6).
COMPUTE R(1,2)=data(1).
COMPUTE R(1,3)=data(2).
COMPUTE R(1,4)=data(3).
COMPUTE R(2,3)=data(4).
COMPUTE R(2,4)=data(5).
COMPUTE R(3,4)=data(6).
PRINT R.
SAVE {I,DET(R)} /OUTFILE='DETS.SAV' /VARIABLES=SAMPLE_ DET.
END LOOP.
END MATRIX.

Any further insight would be appreciated.
Jason

_____________________________________________________________
Jason W. Beckstead, Ph.D.
Associate Professor/Quantitative Methodologist
University of South Florida College of Nursing
12901 Bruce B. Downs Blvd., MDC22, Tampa, FL 33612, USA
Statistical Editor, International Journal of Nursing Studies
phone: +1.813.974.7667 fax: +1.813.974.5418
personal website: http://personal.health.usf.edu/jbeckste/
International Journal of Nursing Studies http://www.elsevier.com/ijns

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

David Marso

Re: saving MATRIX output to SAV file

Administrator

This works.
I misled you with MDIAG. I should have used IDENT.
The : method if assignment is faster than element by element.
Note: also creating just the upper triangle then smashing it into its transpose and adding an Identity matrix
DATA LIST FREE/rYX1 rYX2 rYpdct rX1X2 rX1pdct rX2pdct SAMPLE_.
BEGIN DATA.
.59 .41 .71 -.05 .67 .70 1.00
.49 .35 .62 -.06 .79 .55 2.00
.55 .38 .67 -.06 .78 .56 3.00
END DATA.
MATRIX.
GET data/FILE=*.
LOOP i=1 TO NROW(data).
COMPUTE R=MAKE(4,4,0).
COMPUTE R(1,2:4)=data(i,1:3).
COMPUTE R(2,3:4)=data(i,4:5).
COMPUTE R(4,3)=data(i,6).
COMPUTE R=R+T(R) + IDENT(4).
/*PRINT R.
SAVE {I,DET(R)} /OUTFILE='DETS.SAV' /VARIABLES=SAMPLE_ DET.
END LOOP.
END MATRIX.

David Marso wrote

You want data(I,1) for the first assignment and I in the others as well.

Beckstead, Jason wrote

David and others,
thanks for your suggestions. I tried adapting your code but I must be missing something with the LOOP in MATRIX procedure. I keep getting an undefined error. My file structure is wide, rather than tall so all 6 vars are on the same line for each sample, but I don't think that is the problem. I think it is with the LOOP because I can't even get the code to print 3 R mtxs. Here is the set up for the first 3 of 10,000 samples.

DATA LIST FREE/rYX1 rYX2 rYpdct rX1X2 rX1pdct rX2pdct SAMPLE_.
BEGIN DATA.
.59 .41 .71 -.05 .67 .70 1.00
.49 .35 .62 -.06 .79 .55 2.00
.55 .38 .67 -.06 .78 .56 3.00
END DATA.
MATRIX.
GET data/FILE=*.
LOOP i=1 TO NROW(data).
COMPUTE R=MAKE(4,4,1).
COMPUTE R(2,1)=data(1).
COMPUTE R(3,1)=data(2).
COMPUTE R(4,1)=data(3).
COMPUTE R(3,2)=data(4).
COMPUTE R(4,2)=data(5).
COMPUTE R(4,3)=data(6).
COMPUTE R(1,2)=data(1).
COMPUTE R(1,3)=data(2).
COMPUTE R(1,4)=data(3).
COMPUTE R(2,3)=data(4).
COMPUTE R(2,4)=data(5).
COMPUTE R(3,4)=data(6).
PRINT R.
SAVE {I,DET(R)} /OUTFILE='DETS.SAV' /VARIABLES=SAMPLE_ DET.
END LOOP.
END MATRIX.

Any further insight would be appreciated.
Jason

_____________________________________________________________
Jason W. Beckstead, Ph.D.
Associate Professor/Quantitative Methodologist
University of South Florida College of Nursing
12901 Bruce B. Downs Blvd., MDC22, Tampa, FL 33612, USA
Statistical Editor, International Journal of Nursing Studies
phone: +1.813.974.7667 fax: +1.813.974.5418
personal website: http://personal.health.usf.edu/jbeckste/
International Journal of Nursing Studies http://www.elsevier.com/ijns

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

Kirill Orlov

Re: saving MATRIX output to SAV file

In reply to this post by Beckstead, Jason

Did you mean to do this?

DATA LIST FREE/rYX1 rYX2 rYpdct rX1X2 rX1pdct rX2pdct SAMPLE_.BEGIN DATA. .59 .41 .71 -.05 .67 .70 1.00 .49 .35 .62 -.06 .79 .55 2.00 .55 .38 .67 -.06 .78 .56 3.00END DATA.MATRIX.GET data/FILE=*.LOOP i=1 TO NROW(data).comp R=MAKE(4,4,0).COMPUTE R(2,1)=data(i,1).COMPUTE R(3,1)=data(i,2).COMPUTE R(4,1)=data(i,3).COMPUTE R(3,2)=data(i,4).COMPUTE R(4,2)=data(i,5).COMPUTE R(4,3)=data(i,6).comp r= r+t(r).call setdiag(r,1).PRINT R.SAVE {I,DET(R)} /OUTFILE= * /VARIABLES=SAMPLE_ DET.END LOOP.END MATRIX.

22.02.2015 0:53, Beckstead, Jason пишет:

David and others,
thanks for your suggestions. I tried adapting your code but I must be missing something with the LOOP in MATRIX procedure. I keep getting an undefined error. My file structure is wide, rather than tall so all 6 vars are on the same line for each sample, but I don't think that is the problem. I think it is with the LOOP because I can't even get the code to print 3 R mtxs. Here is the set up for the first 3 of 10,000 samples.

David Marso

Re: saving MATRIX output to SAV file

Administrator

In reply to this post by David Marso

Refinement: UNTESTED.
****.
DATA LIST FREE/rYX1 rYX2 rYpdct rX1X2 rX1pdct rX2pdct SAMPLE_.
BEGIN DATA.
.59 .41 .71 -.05 .67 .70 1.00
.49 .35 .62 -.06 .79 .55 2.00
.55 .38 .67 -.06 .78 .56 3.00
END DATA.
MATRIX.
GET data/FILE=*.
COMPUTE numelem=NCOL(data)-1.
COMPUTE numcol=0.
LOOP k=1 TO 100000.
+ COMPUTE numcol=numcol+k.
END LOOP IF (numcol EQ numelem).
LOOP i=1 TO NROW(data).
+ COMPUTE R=MAKE(numcol,numcol,0).
+ COMPUTE curitem=1.
+ LOOP j=1 TO numcol.
+ COMPUTE j1=j+1.
+ COMPUTE enditem=curitem + numcol - j1.
+ COMPUTE r(j,j1:numcol)=data(i,curitem:enditem).
+ END LOOP.
SAVE {I,DET(R=R+T(R) + IDENT(numcol))} /OUTFILE='DETS.SAV' /VARIABLES=SAMPLE_ DET.
END LOOP.
END MATRIX.
---------------------------------

David Marso wrote

This works.
I misled you with MDIAG. I should have used IDENT.
The : method if assignment is faster than element by element.
Note: also creating just the upper triangle then smashing it into its transpose and adding an Identity matrix
DATA LIST FREE/rYX1 rYX2 rYpdct rX1X2 rX1pdct rX2pdct SAMPLE_.
BEGIN DATA.
.59 .41 .71 -.05 .67 .70 1.00
.49 .35 .62 -.06 .79 .55 2.00
.55 .38 .67 -.06 .78 .56 3.00
END DATA.
MATRIX.
GET data/FILE=*.
LOOP i=1 TO NROW(data).
COMPUTE R=MAKE(4,4,0).
COMPUTE R(1,2:4)=data(i,1:3).
COMPUTE R(2,3:4)=data(i,4:5).
COMPUTE R(4,3)=data(i,6).
COMPUTE R=R+T(R) + IDENT(4).
/*PRINT R.
SAVE {I,DET(R)} /OUTFILE='DETS.SAV' /VARIABLES=SAMPLE_ DET.
END LOOP.
END MATRIX.

David Marso wrote

You want data(I,1) for the first assignment and I in the others as well.

Beckstead, Jason wrote

David and others,
thanks for your suggestions. I tried adapting your code but I must be missing something with the LOOP in MATRIX procedure. I keep getting an undefined error. My file structure is wide, rather than tall so all 6 vars are on the same line for each sample, but I don't think that is the problem. I think it is with the LOOP because I can't even get the code to print 3 R mtxs. Here is the set up for the first 3 of 10,000 samples.

DATA LIST FREE/rYX1 rYX2 rYpdct rX1X2 rX1pdct rX2pdct SAMPLE_.
BEGIN DATA.
.59 .41 .71 -.05 .67 .70 1.00
.49 .35 .62 -.06 .79 .55 2.00
.55 .38 .67 -.06 .78 .56 3.00
END DATA.
MATRIX.
GET data/FILE=*.
LOOP i=1 TO NROW(data).
COMPUTE R=MAKE(4,4,1).
COMPUTE R(2,1)=data(1).
COMPUTE R(3,1)=data(2).
COMPUTE R(4,1)=data(3).
COMPUTE R(3,2)=data(4).
COMPUTE R(4,2)=data(5).
COMPUTE R(4,3)=data(6).
COMPUTE R(1,2)=data(1).
COMPUTE R(1,3)=data(2).
COMPUTE R(1,4)=data(3).
COMPUTE R(2,3)=data(4).
COMPUTE R(2,4)=data(5).
COMPUTE R(3,4)=data(6).
PRINT R.
SAVE {I,DET(R)} /OUTFILE='DETS.SAV' /VARIABLES=SAMPLE_ DET.
END LOOP.
END MATRIX.

Any further insight would be appreciated.
Jason

_____________________________________________________________
Jason W. Beckstead, Ph.D.
Associate Professor/Quantitative Methodologist
University of South Florida College of Nursing
12901 Bruce B. Downs Blvd., MDC22, Tampa, FL 33612, USA
Statistical Editor, International Journal of Nursing Studies
phone: +1.813.974.7667 fax: +1.813.974.5418
personal website: http://personal.health.usf.edu/jbeckste/
International Journal of Nursing Studies http://www.elsevier.com/ijns

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

Kirill Orlov

Re: saving MATRIX output to SAV file

David, it didn't work.
Also, an aside note, call setdiag is much better than adding ident().
Compare elapsed time:

matrix.comp x= make(10000,10000,0).call setdiag(x,1).end matrix.matrix.comp x= make(10000,10000,0).comp x= x+ident(10000).end matrix.

22.02.2015 11:32, David Marso пишет:

Refinement: UNTESTED.
****.
DATA LIST FREE/rYX1 rYX2 rYpdct rX1X2 rX1pdct rX2pdct SAMPLE_. 
BEGIN DATA. 
 .59      .41      .71     -.05      .67      .70     1.00 
 .49      .35      .62     -.06      .79      .55     2.00 
 .55      .38      .67     -.06      .78      .56     3.00 
END DATA. 
MATRIX. 
GET data/FILE=*. 
COMPUTE numelem=NCOL(data)-1.
COMPUTE numcol=0.
LOOP k=1 TO 100000.
+  COMPUTE numcol=numcol+k.
END LOOP IF (numcol EQ numelem).
LOOP i=1 TO NROW(data).
+  COMPUTE R=MAKE(numcol,numcol,0). 
+  COMPUTE curitem=1.
+  LOOP j=1 TO numcol. 
+    COMPUTE j1=j+1.
+    COMPUTE enditem=curitem + numcol - j1.
+    COMPUTE r(j,j1:numcol)=data(i,curitem:enditem).
+  END LOOP.
SAVE {I,DET(R=R+T(R) + IDENT(numcol))} /OUTFILE='DETS.SAV'
/VARIABLES=SAMPLE_ DET. 
END LOOP.
END MATRIX.

Beckstead, Jason

Re: saving MATRIX output to SAV file

In reply to this post by David Marso

David,
Thanks for suggestions. I figured the problem was with the LOOP structure for writing/saving out results, which I fixed. I appreciate your elegant mods to the input structure (a little matrix algebra goes a long way). Here is my final code that did the trick.

*THIS WORKS -------------------------------------------------.
DATA LIST FREE/rYX1 rYX2 rYpdct rX1X2 rX1pdct rX2pdct SAMPLE_.
BEGIN DATA.
.59 .41 .71 -.05 .67 .70 1.00
.49 .35 .62 -.06 .79 .55 2.00
.55 .38 .67 -.06 .78 .56 3.00
END DATA.
MATRIX.
GET data/FILE=*.
COMPUTE D=MAKE(NROW(data),2,0).
LOOP i=1 TO NROW(data).
COMPUTE R=MAKE(4,4,1).
COMPUTE R(2,1)=data(i,1).
COMPUTE R(3,1)=data(i,2).
COMPUTE R(4,1)=data(i,3).
COMPUTE R(3,2)=data(i,4).
COMPUTE R(4,2)=data(i,5).
COMPUTE R(4,3)=data(i,6).
COMPUTE R(1,2)=data(i,1).
COMPUTE R(1,3)=data(i,2).
COMPUTE R(1,4)=data(i,3).
COMPUTE R(2,3)=data(i,4).
COMPUTE R(2,4)=data(i,5).
COMPUTE R(3,4)=data(i,6).
COMPUTE D(i,1:2)={i,DET(R)}.
END LOOP.
SAVE D /OUTFILE='C:\DETS.SAV' /VARIABLES=SAMPLE_ DET.
END MATRIX.

_____________________________________________________________
Jason W. Beckstead, Ph.D.
Associate Professor/Quantitative Methodologist
University of South Florida College of Nursing
12901 Bruce B. Downs Blvd., MDC22, Tampa, FL 33612, USA
Statistical Editor, International Journal of Nursing Studies
phone: +1.813.974.7667 fax: +1.813.974.5418
personal website: http://personal.health.usf.edu/jbeckste/
International Journal of Nursing Studies http://www.elsevier.com/ijns

________________________________________
From: SPSSX(r) Discussion [[hidden email]] on behalf of David Marso [[hidden email]]
Sent: Sunday, February 22, 2015 3:32 AM
To: [hidden email]
Subject: Re: saving MATRIX output to SAV file

Refinement: UNTESTED.
****.
DATA LIST FREE/rYX1 rYX2 rYpdct rX1X2 rX1pdct rX2pdct SAMPLE_.
BEGIN DATA.
.59 .41 .71 -.05 .67 .70 1.00
.49 .35 .62 -.06 .79 .55 2.00
.55 .38 .67 -.06 .78 .56 3.00
END DATA.
MATRIX.
GET data/FILE=*.
COMPUTE numelem=NCOL(data)-1.
COMPUTE numcol=0.
LOOP k=1 TO 100000.
+ COMPUTE numcol=numcol+k.
END LOOP IF (numcol EQ numelem).
LOOP i=1 TO NROW(data).
+ COMPUTE R=MAKE(numcol,numcol,0).
+ COMPUTE curitem=1.
+ LOOP j=1 TO numcol.
+ COMPUTE j1=j+1.
+ COMPUTE enditem=curitem + numcol - j1.
+ COMPUTE r(j,j1:numcol)=data(i,curitem:enditem).
+ END LOOP.
SAVE {I,DET(R=R+T(R) + IDENT(numcol))} /OUTFILE='DETS.SAV'
/VARIABLES=SAMPLE_ DET.
END LOOP.
END MATRIX.
---------------------------------

David Marso wrote

> This works.
> I misled you with MDIAG. I should have used IDENT.
> The : method if assignment is faster than element by element.
> Note: also creating just the upper triangle then smashing it into its
> transpose and adding an Identity matrix
> DATA LIST FREE/rYX1 rYX2 rYpdct rX1X2 rX1pdct rX2pdct SAMPLE_.
> BEGIN DATA.
> .59 .41 .71 -.05 .67 .70 1.00
> .49 .35 .62 -.06 .79 .55 2.00
> .55 .38 .67 -.06 .78 .56 3.00
> END DATA.
> MATRIX.
> GET data/FILE=*.
> LOOP i=1 TO NROW(data).
> COMPUTE R=MAKE(4,4,0).
> COMPUTE R(1,2:4)=data(i,1:3).
> COMPUTE R(2,3:4)=data(i,4:5).
> COMPUTE R(4,3)=data(i,6).
> COMPUTE R=R+T(R) + IDENT(4).
> /*PRINT R.
> SAVE {I,DET(R)} /OUTFILE='DETS.SAV' /VARIABLES=SAMPLE_ DET.
> END LOOP.
> END MATRIX.
> David Marso wrote
>> You want data(I,1) for the first assignment and I in the others as well.
>> Beckstead, Jason wrote
>>> David and others,
>>> thanks for your suggestions. I tried adapting your code but I must be
>>> missing something with the LOOP in MATRIX procedure. I keep getting an
>>> undefined error. My file structure is wide, rather than tall so all 6
>>> vars are on the same line for each sample, but I don't think that is the
>>> problem. I think it is with the LOOP because I can't even get the code
>>> to print 3 R mtxs. Here is the set up for the first 3 of 10,000 samples.
>>>
>>> DATA LIST FREE/rYX1 rYX2 rYpdct rX1X2 rX1pdct rX2pdct SAMPLE_.
>>> BEGIN DATA.
>>> .59 .41 .71 -.05 .67 .70 1.00
>>> .49 .35 .62 -.06 .79 .55 2.00
>>> .55 .38 .67 -.06 .78 .56 3.00
>>> END DATA.
>>> MATRIX.
>>> GET data/FILE=*.
>>> LOOP i=1 TO NROW(data).
>>> COMPUTE R=MAKE(4,4,1).
>>> COMPUTE R(2,1)=data(1).
>>> COMPUTE R(3,1)=data(2).
>>> COMPUTE R(4,1)=data(3).
>>> COMPUTE R(3,2)=data(4).
>>> COMPUTE R(4,2)=data(5).
>>> COMPUTE R(4,3)=data(6).
>>> COMPUTE R(1,2)=data(1).
>>> COMPUTE R(1,3)=data(2).
>>> COMPUTE R(1,4)=data(3).
>>> COMPUTE R(2,3)=data(4).
>>> COMPUTE R(2,4)=data(5).
>>> COMPUTE R(3,4)=data(6).
>>> PRINT R.
>>> SAVE {I,DET(R)} /OUTFILE='DETS.SAV' /VARIABLES=SAMPLE_ DET.
>>> END LOOP.
>>> END MATRIX.
>>>
>>> Any further insight would be appreciated.
>>> Jason
>>>
>>> _____________________________________________________________
>>> Jason W. Beckstead, Ph.D.
>>> Associate Professor/Quantitative Methodologist
>>> University of South Florida College of Nursing
>>> 12901 Bruce B. Downs Blvd., MDC22, Tampa, FL 33612, USA
>>> Statistical Editor, International Journal of Nursing Studies
>>> phone: +1.813.974.7667 fax: +1.813.974.5418
>>> personal website: http://personal.health.usf.edu/jbeckste/
>>> International Journal of Nursing Studies http://www.elsevier.com/ijns
>>>
>>>
>>> =====================
>>> To manage your subscription to SPSSX-L, send a message to

>>> LISTSERV@.UGA

>>> (not to SPSSX-L), with no body text except the
>>> command. To leave the list, send the command
>>> SIGNOFF SPSSX-L
>>> For a list of commands to manage subscriptions, send the command
>>> INFO REFCARD

-----
Please reply to the list and not to my personal email.
Those desiring my consulting or training services please feel free to email me.
---
"Nolite dare sanctum canibus neque mittatis margaritas vestras ante porcos ne forte conculcent eas pedibus suis."
Cum es damnatorum possederunt porcos iens ut salire off sanguinum cliff in abyssum?"
--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/saving-MATRIX-output-to-SAV-file-tp5728761p5728770.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

David Marso

Re: saving MATRIX output to SAV file

Administrator

In reply to this post by David Marso

FIXED and Tested.
Had a few BFs in previous ;-(
--

DATA LIST FREE/rYX1 rYX2 rYpdct rX1X2 rX1pdct rX2pdct SAMPLE_.
BEGIN DATA.
.59 .41 .71 -.05 .67 .70 1.00
.49 .35 .62 -.06 .79 .55 2.00
.55 .38 .67 -.06 .78 .56 3.00
END DATA.

DEFINE VectToUpper (!POS !CHAREND("/") / !POS !TOKENS(1) )
+ COMPUTE curitem=1.
+ COMPUTE NC=NCOL(!2).
+ LOOP j=1 TO NC-1.
+ COMPUTE enditem=curitem + NC - (j+1).
+ COMPUTE !2(j,(j+1): NC) = !1(curitem:enditem).
+ COMPUTE curitem = enditem + 1.
+ END LOOP.
!ENDDEFINE .

MATRIX.
GET data/FILE=*.
COMPUTE numcol=0.
LOOP k=1 TO 100.
+ COMPUTE numcol=numcol+k.
END LOOP IF (numcol EQ NCOL(data)-1).
COMPUTE k=k+1.
LOOP i=1 TO NROW(data).
+ COMPUTE R=MAKE(k,k,0).
+ VectToUpper data(i,:) / R .
+ SAVE {I,DET(R+T(R) + IDENT(k))} /OUTFILE=* /VARIABLES=SAMPLE_ DET.
END LOOP.
END MATRIX.

David Marso wrote

Refinement: UNTESTED.
****.
DATA LIST FREE/rYX1 rYX2 rYpdct rX1X2 rX1pdct rX2pdct SAMPLE_.
BEGIN DATA.
.59 .41 .71 -.05 .67 .70 1.00
.49 .35 .62 -.06 .79 .55 2.00
.55 .38 .67 -.06 .78 .56 3.00
END DATA.
MATRIX.
GET data/FILE=*.
COMPUTE numelem=NCOL(data)-1.
COMPUTE numcol=0.
LOOP k=1 TO 100000.
+ COMPUTE numcol=numcol+k.
END LOOP IF (numcol EQ numelem).
LOOP i=1 TO NROW(data).
+ COMPUTE R=MAKE(numcol,numcol,0).
+ COMPUTE curitem=1.
+ LOOP j=1 TO numcol.
+ COMPUTE j1=j+1.
+ COMPUTE enditem=curitem + numcol - j1.
+ COMPUTE r(j,j1:numcol)=data(i,curitem:enditem).
+ END LOOP.
SAVE {I,DET(R=R+T(R) + IDENT(numcol))} /OUTFILE='DETS.SAV' /VARIABLES=SAMPLE_ DET.
END LOOP.
END MATRIX.
---------------------------------

David Marso wrote

This works.
I misled you with MDIAG. I should have used IDENT.
The : method if assignment is faster than element by element.
Note: also creating just the upper triangle then smashing it into its transpose and adding an Identity matrix
DATA LIST FREE/rYX1 rYX2 rYpdct rX1X2 rX1pdct rX2pdct SAMPLE_.
BEGIN DATA.
.59 .41 .71 -.05 .67 .70 1.00
.49 .35 .62 -.06 .79 .55 2.00
.55 .38 .67 -.06 .78 .56 3.00
END DATA.
MATRIX.
GET data/FILE=*.
LOOP i=1 TO NROW(data).
COMPUTE R=MAKE(4,4,0).
COMPUTE R(1,2:4)=data(i,1:3).
COMPUTE R(2,3:4)=data(i,4:5).
COMPUTE R(4,3)=data(i,6).
COMPUTE R=R+T(R) + IDENT(4).
/*PRINT R.
SAVE {I,DET(R)} /OUTFILE='DETS.SAV' /VARIABLES=SAMPLE_ DET.
END LOOP.
END MATRIX.

David Marso wrote

You want data(I,1) for the first assignment and I in the others as well.

Beckstead, Jason wrote

David and others,
thanks for your suggestions. I tried adapting your code but I must be missing something with the LOOP in MATRIX procedure. I keep getting an undefined error. My file structure is wide, rather than tall so all 6 vars are on the same line for each sample, but I don't think that is the problem. I think it is with the LOOP because I can't even get the code to print 3 R mtxs. Here is the set up for the first 3 of 10,000 samples.

DATA LIST FREE/rYX1 rYX2 rYpdct rX1X2 rX1pdct rX2pdct SAMPLE_.
BEGIN DATA.
.59 .41 .71 -.05 .67 .70 1.00
.49 .35 .62 -.06 .79 .55 2.00
.55 .38 .67 -.06 .78 .56 3.00
END DATA.
MATRIX.
GET data/FILE=*.
LOOP i=1 TO NROW(data).
COMPUTE R=MAKE(4,4,1).
COMPUTE R(2,1)=data(1).
COMPUTE R(3,1)=data(2).
COMPUTE R(4,1)=data(3).
COMPUTE R(3,2)=data(4).
COMPUTE R(4,2)=data(5).
COMPUTE R(4,3)=data(6).
COMPUTE R(1,2)=data(1).
COMPUTE R(1,3)=data(2).
COMPUTE R(1,4)=data(3).
COMPUTE R(2,3)=data(4).
COMPUTE R(2,4)=data(5).
COMPUTE R(3,4)=data(6).
PRINT R.
SAVE {I,DET(R)} /OUTFILE='DETS.SAV' /VARIABLES=SAMPLE_ DET.
END LOOP.
END MATRIX.

Any further insight would be appreciated.
Jason

_____________________________________________________________
Jason W. Beckstead, Ph.D.
Associate Professor/Quantitative Methodologist
University of South Florida College of Nursing
12901 Bruce B. Downs Blvd., MDC22, Tampa, FL 33612, USA
Statistical Editor, International Journal of Nursing Studies
phone: +1.813.974.7667 fax: +1.813.974.5418
personal website: http://personal.health.usf.edu/jbeckste/
International Journal of Nursing Studies http://www.elsevier.com/ijns

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD