SPSSX Discussion

matrix data unscrambling problem

Classic

List

Threaded

14 messages Options

Maguin, Eugene

matrix data unscrambling problem

I have variance-covariance matrix (lower triangle) written to a text file in serial format, five values per line. I need to restructure the data to a lower triangle form so that it can eventually be read into spss via Matrix Data. As I understand what spss can do, there is no way to use spss to read the matrix directly in such way that it is correctly structured. I haven’t been able to think of a way to use spss read the file and restructure the data values into lower triangular form. However, I do figure that it could be done using the Matrix command set, which I’m not familiar with and would prefer to avoid getting familiar. I’m also quite sure it could be done in Python but I’m not familiar with that either.

So any ideas about how to do this using the spss command set?

Thanks, Gene Maguin

Bruce Weaver

Re: matrix data unscrambling problem

Administrator

Gene, maybe it's just a "too late on Friday afternoon" problem, but I'm struggling a bit to understand what the original text file looks like. Can you show us (i.e., paste it into a message)? Thanks.

Maguin, Eugene wrote

I have variance-covariance matrix (lower triangle) written to a text file in serial format, five values per line. I need to restructure the data to a lower triangle form so that it can eventually be read into spss via Matrix Data. As I understand what spss can do, there is no way to use spss to read the matrix directly in such way that it is correctly structured. I haven't been able to think of a way to use spss read the file and restructure the data values into lower triangular form. However, I do figure that it could be done using the Matrix command set, which I'm not familiar with and would prefer to avoid getting familiar. I'm also quite sure it could be done in Python but I'm not familiar with that either.

So any ideas about how to do this using the spss command set?

Thanks, Gene Maguin

--
Bruce Weaver
bweaver@lakeheadu.ca
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

PLEASE NOTE THE FOLLOWING:
1. My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above.
2. The SPSSX Discussion forum on Nabble is no longer linked to the SPSSX-L listserv administered by UGA (https://listserv.uga.edu/).

Maguin, Eugene

Re: matrix data unscrambling problem

Sure.
Suppose you have a 6x6 variance-covariance matrix. Written out in lower triangle (LT) form, you have

11
21 22
31 32 33
41 42 43 44
51 52 53 54 55
61 62 63 64 65 66

Where the numbers are the row,col coordinates. In the real matrix, there would be decimal numbers.
Anyway, the matrix comes out of mplus like this

11 21 22 31 32
33 41 42 43 44
51 52 53 54 55
61 62 63 64 65
66

Five elements per row, six rows total.

So, how to get it back to LT form.

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Bruce Weaver
Sent: Friday, November 15, 2013 4:50 PM
To: [hidden email]
Subject: Re: matrix data unscrambling problem

Gene, maybe it's just a "too late on Friday afternoon" problem, but I'm struggling a bit to understand what the original text file looks like. Can you /show/ us (i.e., paste it into a message)? Thanks.

Maguin, Eugene wrote

> I have variance-covariance matrix (lower triangle) written to a text
> file in serial format, five values per line. I need to restructure the
> data to a lower triangle form so that it can eventually be read into
> spss via Matrix Data. As I understand what spss can do, there is no
> way to use spss to read the matrix directly in such way that it is
> correctly structured. I haven't been able to think of a way to use
> spss read the file and restructure the data values into lower
> triangular form. However, I do figure that it could be done using the
> Matrix command set, which I'm not familiar with and would prefer to
> avoid getting familiar. I'm also quite sure it could be done in Python but I'm not familiar with that either.
>
> So any ideas about how to do this using the spss command set?
>
> Thanks, Gene Maguin

-----
--
Bruce Weaver
[hidden email]
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.

--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/matrix-data-unscrambling-problem-tp5723092p5723094.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

Maguin, Eugene

Re: matrix data unscrambling problem

In reply to this post by Bruce Weaver

I understand how to do it now. Pretty easily actually. Just a little trick. Here's the whole setup. Gene Maguin

* RESTRUCTURE MPLUS SAVEDATA FILE.
* DATA LIST FILE='U:\HILLSIDE\MPLUS2\RESTRUCTURETEST.TXT' FREE / V1 TO V36.
DATA LIST FREE / V1 TO V36.
BEGIN DATA
11 21 22 31 32
33 41 42 43 44
51 52 53 54 55
61 62 63 64 65
66 71 72 73 74
75 76 77 81 82
83 84 85 86 87
88
END DATA.
EXECUTE.
FORMAT V1 TO V36(F2.0).

FLIP. /* right here!.

FORMAT VAR001(F2.0).
COMPUTE ROW=1.
COMPUTE COL=1.
FORMAT ROW COL(F1.0).
DO IF ($CASENUM GE 2).
+ DO IF (LAG(ROW) EQ LAG(COL)).
+ COMPUTE ROW=LAG(ROW)+1.
+ COMPUTE COL=1.
+ ELSE.
+ COMPUTE ROW=LAG(ROW).
+ COMPUTE COL=LAG(COL)+1.
+ END IF.
END IF.
EXECUTE.

DELETE VARIABLES CASE_LBL.

CASESTOVARS ID=ROW/INDEX=COL.

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Bruce Weaver
Sent: Friday, November 15, 2013 4:50 PM
To: [hidden email]
Subject: Re: matrix data unscrambling problem

Gene, maybe it's just a "too late on Friday afternoon" problem, but I'm struggling a bit to understand what the original text file looks like. Can you /show/ us (i.e., paste it into a message)? Thanks.

Maguin, Eugene wrote

Bruce Weaver

Re: matrix data unscrambling problem

Administrator

How about this after the FLIP? ;-)

COMPUTE ROW = TRUNC(VAR001/10).
COMPUTE COL = VAR001 - ROW*10.
FORMATS VAR001 (F2.0) ROW COL (F1).
EXECUTE. /* Needed before DELETE VARIABLES.
DELETE VARIABLES CASE_LBL.
CASESTOVARS ID=ROW/INDEX=COL.

Maguin, Eugene wrote

I understand how to do it now. Pretty easily actually. Just a little trick. Here's the whole setup. Gene Maguin

* RESTRUCTURE MPLUS SAVEDATA FILE.
* DATA LIST FILE='U:\HILLSIDE\MPLUS2\RESTRUCTURETEST.TXT' FREE / V1 TO V36.
DATA LIST FREE / V1 TO V36.
BEGIN DATA
11 21 22 31 32
33 41 42 43 44
51 52 53 54 55
61 62 63 64 65
66 71 72 73 74
75 76 77 81 82
83 84 85 86 87
88
END DATA.
EXECUTE.
FORMAT V1 TO V36(F2.0).

FLIP. /* right here!.

FORMAT VAR001(F2.0).
COMPUTE ROW=1.
COMPUTE COL=1.
FORMAT ROW COL(F1.0).
DO IF ($CASENUM GE 2).
+ DO IF (LAG(ROW) EQ LAG(COL)).
+ COMPUTE ROW=LAG(ROW)+1.
+ COMPUTE COL=1.
+ ELSE.
+ COMPUTE ROW=LAG(ROW).
+ COMPUTE COL=LAG(COL)+1.
+ END IF.
END IF.
EXECUTE.

DELETE VARIABLES CASE_LBL.

CASESTOVARS ID=ROW/INDEX=COL.

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Bruce Weaver
Sent: Friday, November 15, 2013 4:50 PM
To: [hidden email]
Subject: Re: matrix data unscrambling problem

Gene, maybe it's just a "too late on Friday afternoon" problem, but I'm struggling a bit to understand what the original text file looks like. Can you /show/ us (i.e., paste it into a message)? Thanks.

Maguin, Eugene wrote
> I have variance-covariance matrix (lower triangle) written to a text
> file in serial format, five values per line. I need to restructure the
> data to a lower triangle form so that it can eventually be read into
> spss via Matrix Data. As I understand what spss can do, there is no
> way to use spss to read the matrix directly in such way that it is
> correctly structured. I haven't been able to think of a way to use
> spss read the file and restructure the data values into lower
> triangular form. However, I do figure that it could be done using the
> Matrix command set, which I'm not familiar with and would prefer to
> avoid getting familiar. I'm also quite sure it could be done in Python but I'm not familiar with that either.
>
> So any ideas about how to do this using the spss command set?
>
> Thanks, Gene Maguin

-----
--
Bruce Weaver
[hidden email]
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.

--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/matrix-data-unscrambling-problem-tp5723092p5723094.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: matrix data unscrambling problem

Administrator

I don't see the reluctance to learn MATRIX.
It is quite powerful and pretty simple once you get a taste.

DATA LIST FREE / V.
BEGIN DATA
11 21 22 31 32
33 41 42 43 44
51 52 53 54 55
61 62 63 64 65
66
END DATA.
MATRIX.
GET V / FILE * / VARIABLES V.
+ COMPUTE COV=MAKE(6,6,0).
+ COMPUTE #=1.
+ LOOP #R=1 TO 6.
+ LOOP #C=1 TO #R.
+ COMPUTE COV(#R,#C)=V(#).
+ COMPUTE COV(#C,#R)=V(#).
+ COMPUTE #=#+1.
+ END LOOP.
+ END LOOP.
PRINT COV.
END MATRIX.

Run MATRIX procedure:

COV
11 21 31 41 51 61
21 22 32 42 52 62
31 32 33 43 53 63
41 42 43 44 54 64
51 52 53 54 55 65
61 62 63 64 65 66

------ END MATRIX -----

DATA LIST FREE / V.
BEGIN DATA
11 21 22 31 32
33 41 42 43 44
51 52 53 54 55
61 62 63 64 65
66
END DATA.
MATRIX.
GET V / FILE * / VARIABLES V.
+ COMPUTE COV=MAKE(6,6,0).
+ COMPUTE #=1.
+ LOOP #R=1 TO 6.
+ LOOP #C=1 TO #R.
+ COMPUTE COV(#R,#C)=V(#).
+ COMPUTE #=#+1.
+ END LOOP.
+ END LOOP.
PRINT COV.
END MATRIX.

Run MATRIX procedure:

COV
11 0 0 0 0 0
21 22 0 0 0 0
31 32 33 0 0 0
41 42 43 44 0 0
51 52 53 54 55 0
61 62 63 64 65 66

------ END MATRIX -----

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?"

David Marso

Re: matrix data unscrambling problem

Administrator

Thinking it might be awkward to hard code the size, here is a QAD way to dynamically determine the required size.
DATA LIST FREE / V.
BEGIN DATA
11 21 22 31 32
33 41 42 43 44
51 52 53 54 55
61 62 63 64 65
66
END DATA.
MATRIX.

GET V / FILE * / VARIABLES V.
COMPUTE #=1.
COMPUTE Nelem=NROW(V).
LOOP.
+ COMPUTE Nelem=Nelem-#.
+ COMPUTE #=#+1.
END LOOP IF Nelem LE 0.
COMPUTE Nrc=#-1.

+ COMPUTE COV=MAKE(Nrc,Nrc,0).
+ COMPUTE #=1.
+ LOOP #R=1 TO Nrc.
+ LOOP #C=1 TO #R.
+ COMPUTE COV(#R,#C)=V(#).
+ COMPUTE COV(#C,#R)=V(#).
+ COMPUTE #=#+1.
+ END LOOP.
+ END LOOP.
PRINT COV.
END MATRIX.

Run MATRIX procedure:

COV
11 21 31 41 51 61
21 22 32 42 52 62
31 32 33 43 53 63
41 42 43 44 54 64
51 52 53 54 55 65
61 62 63 64 65 66

------ END MATRIX -----

David Marso wrote

I don't see the reluctance to learn MATRIX.
It is quite powerful and pretty simple once you get a taste.

DATA LIST FREE / V.
BEGIN DATA
11 21 22 31 32
33 41 42 43 44
51 52 53 54 55
61 62 63 64 65
66
END DATA.
MATRIX.
GET V / FILE * / VARIABLES V.
+ COMPUTE COV=MAKE(6,6,0).
+ COMPUTE #=1.
+ LOOP #R=1 TO 6.
+ LOOP #C=1 TO #R.
+ COMPUTE COV(#R,#C)=V(#).
+ COMPUTE COV(#C,#R)=V(#).
+ COMPUTE #=#+1.
+ END LOOP.
+ END LOOP.
PRINT COV.
END MATRIX.

Run MATRIX procedure:

COV
11 21 31 41 51 61
21 22 32 42 52 62
31 32 33 43 53 63
41 42 43 44 54 64
51 52 53 54 55 65
61 62 63 64 65 66

------ END MATRIX -----

DATA LIST FREE / V.
BEGIN DATA
11 21 22 31 32
33 41 42 43 44
51 52 53 54 55
61 62 63 64 65
66
END DATA.
MATRIX.
GET V / FILE * / VARIABLES V.
+ COMPUTE COV=MAKE(6,6,0).
+ COMPUTE #=1.
+ LOOP #R=1 TO 6.
+ LOOP #C=1 TO #R.
+ COMPUTE COV(#R,#C)=V(#).
+ COMPUTE #=#+1.
+ END LOOP.
+ END LOOP.
PRINT COV.
END MATRIX.

Run MATRIX procedure:

COV
11 0 0 0 0 0
21 22 0 0 0 0
31 32 33 0 0 0
41 42 43 44 0 0
51 52 53 54 55 0
61 62 63 64 65 66

------ END MATRIX -----

David Marso

Re: matrix data unscrambling problem

Administrator

Of course all 3 of us missed the obvious ;-(
BTW: SPSS doesn't really have any use for a lower triangular matrix that I can think of.
I assume you are dumping this into the CANCORR macro (per recent posts).
Follow this with MCONVERT and then stick it into CanCorr after clobbering the lines which create the CORRELATION matrix .

MATRIX DATA VAR v1 to v6 /FORMAT FREE /CONTENTS COV.
BEGIN DATA
11 21 22 31 32
33 41 42 43 44
51 52 53 54 55
61 62 63 64 65
66
END DATA.
LIST.

ROWTYPE_ VARNAME_ v1 v2 v3 v4 v5 v6

COV v1 11.0000 21.0000 31.0000 41.0000 51.0000 61.0000
COV v2 21.0000 22.0000 32.0000 42.0000 52.0000 62.0000
COV v3 31.0000 32.0000 33.0000 43.0000 53.0000 63.0000
COV v4 41.0000 42.0000 43.0000 44.0000 54.0000 64.0000
COV v5 51.0000 52.0000 53.0000 54.0000 55.0000 65.0000
COV v6 61.0000 62.0000 63.0000 64.0000 65.0000 66.0000

Number of cases read: 6 Number of cases listed: 6

David Marso wrote

Thinking it might be awkward to hard code the size, here is a QAD way to dynamically determine the required size.
DATA LIST FREE / V.
BEGIN DATA
11 21 22 31 32
33 41 42 43 44
51 52 53 54 55
61 62 63 64 65
66
END DATA.
MATRIX.

GET V / FILE * / VARIABLES V.
COMPUTE #=1.
COMPUTE Nelem=NROW(V).
LOOP.
+ COMPUTE Nelem=Nelem-#.
+ COMPUTE #=#+1.
END LOOP IF Nelem LE 0.
COMPUTE Nrc=#-1.

+ COMPUTE COV=MAKE(Nrc,Nrc,0).
+ COMPUTE #=1.
+ LOOP #R=1 TO Nrc.
+ LOOP #C=1 TO #R.
+ COMPUTE COV(#R,#C)=V(#).
+ COMPUTE COV(#C,#R)=V(#).
+ COMPUTE #=#+1.
+ END LOOP.
+ END LOOP.
PRINT COV.
END MATRIX.

Run MATRIX procedure:

COV
11 21 31 41 51 61
21 22 32 42 52 62
31 32 33 43 53 63
41 42 43 44 54 64
51 52 53 54 55 65
61 62 63 64 65 66

------ END MATRIX -----

David Marso wrote

I don't see the reluctance to learn MATRIX.
It is quite powerful and pretty simple once you get a taste.

DATA LIST FREE / V.
BEGIN DATA
11 21 22 31 32
33 41 42 43 44
51 52 53 54 55
61 62 63 64 65
66
END DATA.
MATRIX.
GET V / FILE * / VARIABLES V.
+ COMPUTE COV=MAKE(6,6,0).
+ COMPUTE #=1.
+ LOOP #R=1 TO 6.
+ LOOP #C=1 TO #R.
+ COMPUTE COV(#R,#C)=V(#).
+ COMPUTE COV(#C,#R)=V(#).
+ COMPUTE #=#+1.
+ END LOOP.
+ END LOOP.
PRINT COV.
END MATRIX.

Run MATRIX procedure:

COV
11 21 31 41 51 61
21 22 32 42 52 62
31 32 33 43 53 63
41 42 43 44 54 64
51 52 53 54 55 65
61 62 63 64 65 66

------ END MATRIX -----

DATA LIST FREE / V.
BEGIN DATA
11 21 22 31 32
33 41 42 43 44
51 52 53 54 55
61 62 63 64 65
66
END DATA.
MATRIX.
GET V / FILE * / VARIABLES V.
+ COMPUTE COV=MAKE(6,6,0).
+ COMPUTE #=1.
+ LOOP #R=1 TO 6.
+ LOOP #C=1 TO #R.
+ COMPUTE COV(#R,#C)=V(#).
+ COMPUTE #=#+1.
+ END LOOP.
+ END LOOP.
PRINT COV.
END MATRIX.

Run MATRIX procedure:

COV
11 0 0 0 0 0
21 22 0 0 0 0
31 32 33 0 0 0
41 42 43 44 0 0
51 52 53 54 55 0
61 62 63 64 65 66

------ END MATRIX -----

Art Kendall

Re: matrix data unscrambling problem

In reply to this post by Maguin, Eugene

does mplus have another way of outputting the matrix?

Art Kendall
Social Research Consultants

On 11/15/2013 5:07 PM, Maguin, Eugene [via SPSSX Discussion] wrote:

Sure.
Suppose you have a 6x6 variance-covariance matrix. Written out in lower triangle (LT) form, you have

11
21 22
31 32 33
41 42 43 44
51 52 53 54 55
61 62 63 64 65 66

Where the numbers are the row,col coordinates. In the real matrix, there would be decimal numbers.
Anyway, the matrix comes out of mplus like this

11 21 22 31 32
33 41 42 43 44
51 52 53 54 55
61 62 63 64 65
66

Five elements per row, six rows total.

So, how to get it back to LT form.

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Bruce Weaver
Sent: Friday, November 15, 2013 4:50 PM
To: [hidden email]
Subject: Re: matrix data unscrambling problem

Gene, maybe it's just a "too late on Friday afternoon" problem, but I'm struggling a bit to understand what the original text file looks like. Can you /show/ us (i.e., paste it into a message)? Thanks.

Maguin, Eugene wrote

> I have variance-covariance matrix (lower triangle) written to a text
> file in serial format, five values per line. I need to restructure the
> data to a lower triangle form so that it can eventually be read into
> spss via Matrix Data. As I understand what spss can do, there is no
> way to use spss to read the matrix directly in such way that it is
> correctly structured. I haven't been able to think of a way to use
> spss read the file and restructure the data values into lower
> triangular form. However, I do figure that it could be done using the
> Matrix command set, which I'm not familiar with and would prefer to
> avoid getting familiar. I'm also quite sure it could be done in Python but I'm not familiar with that either.
>
> So any ideas about how to do this using the spss command set?
>
> Thanks, Gene Maguin

-----
--
Bruce Weaver
[hidden email]
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.

--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/matrix-data-unscrambling-problem-tp5723092p5723094.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

If you reply to this email, your message will be added to the discussion below:
http://spssx-discussion.1045642.n5.nabble.com/matrix-data-unscrambling-problem-tp5723092p5723095.html

To start a new topic under SPSSX Discussion, email [hidden email]
To unsubscribe from SPSSX Discussion, click here.
NAML

Art Kendall
Social Research Consultants

Bruce Weaver

Re: matrix data unscrambling problem

Administrator

In reply to this post by David Marso

Ah yes...I like that one!

David Marso wrote

Of course all 3 of us missed the obvious ;-(
BTW: SPSS doesn't really have any use for a lower triangular matrix that I can think of.
I assume you are dumping this into the CANCORR macro (per recent posts).
Follow this with MCONVERT and then stick it into CanCorr after clobbering the lines which create the CORRELATION matrix .

MATRIX DATA VAR v1 to v6 /FORMAT FREE /CONTENTS COV.
BEGIN DATA
11 21 22 31 32
33 41 42 43 44
51 52 53 54 55
61 62 63 64 65
66
END DATA.
LIST.

ROWTYPE_ VARNAME_ v1 v2 v3 v4 v5 v6

COV v1 11.0000 21.0000 31.0000 41.0000 51.0000 61.0000
COV v2 21.0000 22.0000 32.0000 42.0000 52.0000 62.0000
COV v3 31.0000 32.0000 33.0000 43.0000 53.0000 63.0000
COV v4 41.0000 42.0000 43.0000 44.0000 54.0000 64.0000
COV v5 51.0000 52.0000 53.0000 54.0000 55.0000 65.0000
COV v6 61.0000 62.0000 63.0000 64.0000 65.0000 66.0000

Number of cases read: 6 Number of cases listed: 6

David Marso wrote

Thinking it might be awkward to hard code the size, here is a QAD way to dynamically determine the required size.
DATA LIST FREE / V.
BEGIN DATA
11 21 22 31 32
33 41 42 43 44
51 52 53 54 55
61 62 63 64 65
66
END DATA.
MATRIX.

GET V / FILE * / VARIABLES V.
COMPUTE #=1.
COMPUTE Nelem=NROW(V).
LOOP.
+ COMPUTE Nelem=Nelem-#.
+ COMPUTE #=#+1.
END LOOP IF Nelem LE 0.
COMPUTE Nrc=#-1.

+ COMPUTE COV=MAKE(Nrc,Nrc,0).
+ COMPUTE #=1.
+ LOOP #R=1 TO Nrc.
+ LOOP #C=1 TO #R.
+ COMPUTE COV(#R,#C)=V(#).
+ COMPUTE COV(#C,#R)=V(#).
+ COMPUTE #=#+1.
+ END LOOP.
+ END LOOP.
PRINT COV.
END MATRIX.

Run MATRIX procedure:

COV
11 21 31 41 51 61
21 22 32 42 52 62
31 32 33 43 53 63
41 42 43 44 54 64
51 52 53 54 55 65
61 62 63 64 65 66

------ END MATRIX -----

David Marso wrote

I don't see the reluctance to learn MATRIX.
It is quite powerful and pretty simple once you get a taste.

DATA LIST FREE / V.
BEGIN DATA
11 21 22 31 32
33 41 42 43 44
51 52 53 54 55
61 62 63 64 65
66
END DATA.
MATRIX.
GET V / FILE * / VARIABLES V.
+ COMPUTE COV=MAKE(6,6,0).
+ COMPUTE #=1.
+ LOOP #R=1 TO 6.
+ LOOP #C=1 TO #R.
+ COMPUTE COV(#R,#C)=V(#).
+ COMPUTE COV(#C,#R)=V(#).
+ COMPUTE #=#+1.
+ END LOOP.
+ END LOOP.
PRINT COV.
END MATRIX.

Run MATRIX procedure:

COV
11 21 31 41 51 61
21 22 32 42 52 62
31 32 33 43 53 63
41 42 43 44 54 64
51 52 53 54 55 65
61 62 63 64 65 66

------ END MATRIX -----

DATA LIST FREE / V.
BEGIN DATA
11 21 22 31 32
33 41 42 43 44
51 52 53 54 55
61 62 63 64 65
66
END DATA.
MATRIX.
GET V / FILE * / VARIABLES V.
+ COMPUTE COV=MAKE(6,6,0).
+ COMPUTE #=1.
+ LOOP #R=1 TO 6.
+ LOOP #C=1 TO #R.
+ COMPUTE COV(#R,#C)=V(#).
+ COMPUTE #=#+1.
+ END LOOP.
+ END LOOP.
PRINT COV.
END MATRIX.

Run MATRIX procedure:

COV
11 0 0 0 0 0
21 22 0 0 0 0
31 32 33 0 0 0
41 42 43 44 0 0
51 52 53 54 55 0
61 62 63 64 65 66

------ END MATRIX -----

David Marso

Re: matrix data unscrambling problem

Administrator

In reply to this post by David Marso

Oops, I meant Ridge Reg

David Marso wrote

Of course all 3 of us missed the obvious ;-(
BTW: SPSS doesn't really have any use for a lower triangular matrix that I can think of.
I assume you are dumping this into the CANCORR macro (per recent posts).
Follow this with MCONVERT and then stick it into CanCorr after clobbering the lines which create the CORRELATION matrix .

MATRIX DATA VAR v1 to v6 /FORMAT FREE /CONTENTS COV.
BEGIN DATA
11 21 22 31 32
33 41 42 43 44
51 52 53 54 55
61 62 63 64 65
66
END DATA.
LIST.

ROWTYPE_ VARNAME_ v1 v2 v3 v4 v5 v6

COV v1 11.0000 21.0000 31.0000 41.0000 51.0000 61.0000
COV v2 21.0000 22.0000 32.0000 42.0000 52.0000 62.0000
COV v3 31.0000 32.0000 33.0000 43.0000 53.0000 63.0000
COV v4 41.0000 42.0000 43.0000 44.0000 54.0000 64.0000
COV v5 51.0000 52.0000 53.0000 54.0000 55.0000 65.0000
COV v6 61.0000 62.0000 63.0000 64.0000 65.0000 66.0000

Number of cases read: 6 Number of cases listed: 6

David Marso wrote

Thinking it might be awkward to hard code the size, here is a QAD way to dynamically determine the required size.
DATA LIST FREE / V.
BEGIN DATA
11 21 22 31 32
33 41 42 43 44
51 52 53 54 55
61 62 63 64 65
66
END DATA.
MATRIX.

GET V / FILE * / VARIABLES V.
COMPUTE #=1.
COMPUTE Nelem=NROW(V).
LOOP.
+ COMPUTE Nelem=Nelem-#.
+ COMPUTE #=#+1.
END LOOP IF Nelem LE 0.
COMPUTE Nrc=#-1.

+ COMPUTE COV=MAKE(Nrc,Nrc,0).
+ COMPUTE #=1.
+ LOOP #R=1 TO Nrc.
+ LOOP #C=1 TO #R.
+ COMPUTE COV(#R,#C)=V(#).
+ COMPUTE COV(#C,#R)=V(#).
+ COMPUTE #=#+1.
+ END LOOP.
+ END LOOP.
PRINT COV.
END MATRIX.

Run MATRIX procedure:

COV
11 21 31 41 51 61
21 22 32 42 52 62
31 32 33 43 53 63
41 42 43 44 54 64
51 52 53 54 55 65
61 62 63 64 65 66

------ END MATRIX -----

David Marso wrote

I don't see the reluctance to learn MATRIX.
It is quite powerful and pretty simple once you get a taste.

DATA LIST FREE / V.
BEGIN DATA
11 21 22 31 32
33 41 42 43 44
51 52 53 54 55
61 62 63 64 65
66
END DATA.
MATRIX.
GET V / FILE * / VARIABLES V.
+ COMPUTE COV=MAKE(6,6,0).
+ COMPUTE #=1.
+ LOOP #R=1 TO 6.
+ LOOP #C=1 TO #R.
+ COMPUTE COV(#R,#C)=V(#).
+ COMPUTE COV(#C,#R)=V(#).
+ COMPUTE #=#+1.
+ END LOOP.
+ END LOOP.
PRINT COV.
END MATRIX.

Run MATRIX procedure:

COV
11 21 31 41 51 61
21 22 32 42 52 62
31 32 33 43 53 63
41 42 43 44 54 64
51 52 53 54 55 65
61 62 63 64 65 66

------ END MATRIX -----

DATA LIST FREE / V.
BEGIN DATA
11 21 22 31 32
33 41 42 43 44
51 52 53 54 55
61 62 63 64 65
66
END DATA.
MATRIX.
GET V / FILE * / VARIABLES V.
+ COMPUTE COV=MAKE(6,6,0).
+ COMPUTE #=1.
+ LOOP #R=1 TO 6.
+ LOOP #C=1 TO #R.
+ COMPUTE COV(#R,#C)=V(#).
+ COMPUTE #=#+1.
+ END LOOP.
+ END LOOP.
PRINT COV.
END MATRIX.

Run MATRIX procedure:

COV
11 0 0 0 0 0
21 22 0 0 0 0
31 32 33 0 0 0
41 42 43 44 0 0
51 52 53 54 55 0
61 62 63 64 65 66

------ END MATRIX -----

Maguin, Eugene

Re: matrix data unscrambling problem

In reply to this post by David Marso

Thanks for all your replies.
Bruce: In my example, I used the (I,j) values as x(I,j) values to keep track of operation. The real matrices are just covariance matrices.

David: It's more easily thought of as a matrix problem, I agree.

Art: I think the answer is No. Matrices are written to the output (listing) file in lower triangular but only with three decimal values.

Gene Maguin

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of David Marso
Sent: Friday, November 15, 2013 6:02 PM
To: [hidden email]
Subject: Re: matrix data unscrambling problem

I don't see the reluctance to learn MATRIX.
It is quite powerful and pretty simple once you get a taste.

DATA LIST FREE / V.
BEGIN DATA
11 21 22 31 32
33 41 42 43 44
51 52 53 54 55
61 62 63 64 65
66
END DATA.
MATRIX.
GET V / FILE * / VARIABLES V.
+ COMPUTE COV=MAKE(6,6,0).
+ COMPUTE #=1.
+ LOOP #R=1 TO 6.
+ LOOP #C=1 TO #R.
+ COMPUTE COV(#R,#C)=V(#).
+ COMPUTE COV(#C,#R)=V(#).
+ COMPUTE #=#+1.
+ END LOOP.
+ END LOOP.
PRINT COV.
END MATRIX.

Run MATRIX procedure:

COV
11 21 31 41 51 61
21 22 32 42 52 62
31 32 33 43 53 63
41 42 43 44 54 64
51 52 53 54 55 65
61 62 63 64 65 66

------ END MATRIX -----

DATA LIST FREE / V.
BEGIN DATA
11 21 22 31 32
33 41 42 43 44
51 52 53 54 55
61 62 63 64 65
66
END DATA.
MATRIX.
GET V / FILE * / VARIABLES V.
+ COMPUTE COV=MAKE(6,6,0).
+ COMPUTE #=1.
+ LOOP #R=1 TO 6.
+ LOOP #C=1 TO #R.
+ COMPUTE COV(#R,#C)=V(#).
+ COMPUTE #=#+1.
+ END LOOP.
+ END LOOP.
PRINT COV.
END MATRIX.

Run MATRIX procedure:

COV
11 0 0 0 0 0
21 22 0 0 0 0
31 32 33 0 0 0
41 42 43 44 0 0
51 52 53 54 55 0
61 62 63 64 65 66

------ END MATRIX -----

-----
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/matrix-data-unscrambling-problem-tp5723092p5723099.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: matrix data unscrambling problem

Administrator

Gene,
In a post later in this thread I demonstrated that this can be read directly using MATRIX DATA
Only qualifiers are to use FORMAT=FREE and CONTENTS=COV (since default is LIST and CORR).
To push this into RidgeReg you will need to convert it to a correlation matrix and do some data manipulations to put the N's and Means into appropriate rows (alternatively modify the macro).
Here I demonstrate the issues.
DATA LIST FREE / x1 TO X4 .
BEGIN DATA
5 2 5 1 2 5 6 1 5 6 5 1 6 3 5 1 2 3 5 6 2 5
3 6 2 1 5 6 1 2 5 3 5 1 3 5 5 3 5 1 3 6 1 5
3 5 1 2 5 3 6 1 5 3 1 5 3 7 5 1 7 3 1 5 3 1
END DATA.
CORR VAR x1 to x4 /MISSING = LISTWISE/ MATRIX OUT (*).
LIST.

MATRIX FILE FORMAT ASSUMED BY RIDGEREG.
ROWTYPE_ VARNAME_ x1 x2 x3 x4

MEAN 3.8125000 3.6250000 3.8750000 3.1250000
STDDEV 1.7594980 1.8211718 1.8929694 2.1870833
N 16.0000000 16.0000000 16.0000000 16.0000000
CORR x1 1.0000000 -.1690411 -.2076655 -.2533673
CORR x2 -.1690411 1.0000000 -.1305323 -.2887231
CORR x3 -.2076655 -.1305323 1.0000000 -.5112620
CORR x4 -.2533673 -.2887231 -.5112620 1.0000000

Number of cases read: 7 Number of cases listed: 7

MCONVERT.
LIST.

ROWTYPE_ VARNAME_ x1 x2 x3 x4

MEAN 3.8125000 3.6250000 3.8750000 3.1250000
STDDEV 1.7594980 1.8211718 1.8929694 2.1870833
N 16.0000000 16.0000000 16.0000000 16.0000000
COV x1 3.0958333 -.5416667 -.6916667 -.9750000
COV x2 -.5416667 3.3166667 -.4500000 -1.1500000
COV x3 -.6916667 -.4500000 3.5833333 -2.1166667
COV x4 -.9750000 -1.1500000 -2.1166667 4.7833333

WHAT YOU WOULD BE PROVIDING AS INPUT
MATRIX DATA VAR x1 TO x4
/ FORMAT FREE
/ CONTENTS Mean N_VECTOR COV .
BEGIN DATA
3.8125000 3.6250000 3.8750000 3.1250000
16.0000000 16.0000000 16.0000000 16.0000000
3.0958333 -.5416667 3.3166667 -.6916667 -.4500000 3.5833333 -.9750000 -1.1500000 -2.1166667 4.7833333
END DATA.
LIST.

OOPS! This does NOT agree with what RidgeReg assumes !
ROWTYPE_ VARNAME_ x1 x2 x3 x4

N 16.0000 16.0000 16.0000 16.0000
MEAN 3.8125 3.6250 3.8750 3.1250
COV x1 3.0958 -.5417 -.6917 -.9750
COV x2 -.5417 3.3167 -.4500 -1.1500
COV x3 -.6917 -.4500 3.5833 -2.1167
COV x4 -.9750 -1.1500 -2.1167 4.7833

Number of cases read: 6 Number of cases listed: 6

MCONVERT.
LIST.

NEITHER DOES THIS
ROWTYPE_ VARNAME_ x1 x2 x3 x4

N 16.0000000 16.0000000 16.0000000 16.0000000
MEAN 3.8125000 3.6250000 3.8750000 3.1250000
STDDEV 1.7594980 1.8211718 1.8929694 2.1870833
CORR x1 1.0000000 -.1690411 -.2076656 -.2533673
CORR x2 -.1690411 1.0000000 -.1305323 -.2887231
CORR x3 -.2076656 -.1305323 1.0000000 -.5112620
CORR x4 -.2533673 -.2887231 -.5112620 1.0000000

Number of cases read: 7 Number of cases listed: 7

Notice that this produces a matrix file which is NOT identical to that created by CORRELATIONS ;-(((
the RidgeReg Macro (RR) references the N from row 3 (where here they occur in row 1)
compute n=x(3,1).
needs to be changed to
compute n=x(1,1).

RR assumes the Mean vector is in row 1, where here they are in row 2.
RR assumes the SDs are row 2 where here the end up in Row 3.
compute xmean=x(1,2:(nv+1)).
compute ybar=x(1,1).
compute std=t(x(2,2:(nv+1))).
compute sy=x(2,1).

NEEDS TO BE CHANGED TO
compute xmean=x(2,2:(nv+1)).
compute ybar=x(3,1).
compute std=t(x(3,2:(nv+1))).
compute sy=x(3,1).

You will also need to comment out the raw data read in the beginning.

correlations variables=!dep !enter
/missing=listwise
/matrix out(corr).

HTH, David

Maguin, Eugene wrote

Thanks for all your replies.
Bruce: In my example, I used the (I,j) values as x(I,j) values to keep track of operation. The real matrices are just covariance matrices.

David: It's more easily thought of as a matrix problem, I agree.

Art: I think the answer is No. Matrices are written to the output (listing) file in lower triangular but only with three decimal values.

Gene Maguin

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of David Marso
Sent: Friday, November 15, 2013 6:02 PM
To: [hidden email]
Subject: Re: matrix data unscrambling problem

I don't see the reluctance to learn MATRIX.
It is quite powerful and pretty simple once you get a taste.

DATA LIST FREE / V.
BEGIN DATA
11 21 22 31 32
33 41 42 43 44
51 52 53 54 55
61 62 63 64 65
66
END DATA.
MATRIX.
GET V / FILE * / VARIABLES V.
+ COMPUTE COV=MAKE(6,6,0).
+ COMPUTE #=1.
+ LOOP #R=1 TO 6.
+ LOOP #C=1 TO #R.
+ COMPUTE COV(#R,#C)=V(#).
+ COMPUTE COV(#C,#R)=V(#).
+ COMPUTE #=#+1.
+ END LOOP.
+ END LOOP.
PRINT COV.
END MATRIX.

Run MATRIX procedure:

COV
11 21 31 41 51 61
21 22 32 42 52 62
31 32 33 43 53 63
41 42 43 44 54 64
51 52 53 54 55 65
61 62 63 64 65 66

------ END MATRIX -----

DATA LIST FREE / V.
BEGIN DATA
11 21 22 31 32
33 41 42 43 44
51 52 53 54 55
61 62 63 64 65
66
END DATA.
MATRIX.
GET V / FILE * / VARIABLES V.
+ COMPUTE COV=MAKE(6,6,0).
+ COMPUTE #=1.
+ LOOP #R=1 TO 6.
+ LOOP #C=1 TO #R.
+ COMPUTE COV(#R,#C)=V(#).
+ COMPUTE #=#+1.
+ END LOOP.
+ END LOOP.
PRINT COV.
END MATRIX.

Run MATRIX procedure:

COV
11 0 0 0 0 0
21 22 0 0 0 0
31 32 33 0 0 0
41 42 43 44 0 0
51 52 53 54 55 0
61 62 63 64 65 66

------ END MATRIX -----

-----
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/matrix-data-unscrambling-problem-tp5723092p5723099.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

Maguin, Eugene

Re: matrix data unscrambling problem

Thanks. Yes, that is what I'm going do. I've tested that part out. Turns out that RidgeReg has more flexibility than I originally thought because it will run through a range of k values or will execute for a specific k value. The output depends on the choice of range or specific. Also turns out that the input requirements differ for the choice. If the computation is for a specific value, means are required. If a range of values, then means aren't required. Turns out that mplus writes the EM covariance matrix in E format but the means and N are not included in the save file. They are, however, are printed, albeit in a different format. Although I think only the range of k's option will be picked; flexibility is better.

RidgeReg is set up to save the data file in a temp file and compute the correlations within the macro. Instead, my idea is, as you said,
to use matrix data to read the cov matrix and mconvert to convert it to corrs but outside of the macro , which means I'll comment out some lines in the macro. Originally, I thought of putting matrix data in the macro but I'd need to alter the macro call a bit to pass the N value. Easier, I think, to do so outside the macro as this is a one-time project. However, the drawback to this plan is the need to insert the rowtype variable and assign it values. Your point is that the get command in matrix would avoid this drawback, and it would, I think. It'd interesting to do this because I'd certainly learn something. The tradeoff is time to get minimal proficiency.

Gene Maguin

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of David Marso
Sent: Monday, November 18, 2013 10:47 AM
To: [hidden email]
Subject: Re: matrix data unscrambling problem

Gene,
In a post later in this thread I demonstrated that this can be read directly using MATRIX DATA Only qualifiers are to use FORMAT=FREE and CONTENTS=COV (since default is LIST and CORR).
To push this into RidgeReg you will need to convert it to a correlation matrix and do some data manipulations to put the N's and Means into appropriate rows (alternatively modify the macro).
Here I demonstrate the issues.
DATA LIST FREE / x1 TO X4 .
BEGIN DATA
5 2 5 1 2 5 6 1 5 6 5 1 6 3 5 1 2 3 5 6 2 5
3 6 2 1 5 6 1 2 5 3 5 1 3 5 5 3 5 1 3 6 1 5
3 5 1 2 5 3 6 1 5 3 1 5 3 7 5 1 7 3 1 5 3 1 END DATA.
CORR VAR x1 to x4 /MISSING = LISTWISE/ MATRIX OUT (*).
LIST.

*MATRIX FILE FORMAT ASSUMED BY RIDGEREG. *
ROWTYPE_ VARNAME_ x1 x2 x3 x4

MEAN 3.8125000 3.6250000 3.8750000 3.1250000
STDDEV 1.7594980 1.8211718 1.8929694 2.1870833
N 16.0000000 16.0000000 16.0000000 16.0000000
CORR x1 1.0000000 -.1690411 -.2076655 -.2533673
CORR x2 -.1690411 1.0000000 -.1305323 -.2887231
CORR x3 -.2076655 -.1305323 1.0000000 -.5112620
CORR x4 -.2533673 -.2887231 -.5112620 1.0000000

Number of cases read: 7 Number of cases listed: 7

MCONVERT.
LIST.

ROWTYPE_ VARNAME_ x1 x2 x3 x4

MEAN 3.8125000 3.6250000 3.8750000 3.1250000
STDDEV 1.7594980 1.8211718 1.8929694 2.1870833
N 16.0000000 16.0000000 16.0000000 16.0000000
COV x1 3.0958333 -.5416667 -.6916667 -.9750000
COV x2 -.5416667 3.3166667 -.4500000 -1.1500000
COV x3 -.6916667 -.4500000 3.5833333 -2.1166667
COV x4 -.9750000 -1.1500000 -2.1166667 4.7833333

*WHAT YOU WOULD BE PROVIDING AS INPUT*
MATRIX DATA VAR x1 TO x4
/ FORMAT FREE
/ CONTENTS Mean N_VECTOR COV .
BEGIN DATA
3.8125000 3.6250000 3.8750000 3.1250000
16.0000000 16.0000000 16.0000000 16.0000000
3.0958333 -.5416667 3.3166667 -.6916667 -.4500000 3.5833333 -.9750000
-1.1500000 -2.1166667 4.7833333
END DATA.
LIST.

*OOPS! This does NOT agree with what RidgeReg assumes !
*ROWTYPE_ VARNAME_ x1 x2 x3 x4

N 16.0000 16.0000 16.0000 16.0000
MEAN 3.8125 3.6250 3.8750 3.1250
COV x1 3.0958 -.5417 -.6917 -.9750
COV x2 -.5417 3.3167 -.4500 -1.1500
COV x3 -.6917 -.4500 3.5833 -2.1167
COV x4 -.9750 -1.1500 -2.1167 4.7833

Number of cases read: 6 Number of cases listed: 6

*MCONVERT.*
LIST.

*NEITHER DOES THIS*
ROWTYPE_ VARNAME_ x1 x2 x3 x4

N 16.0000000 16.0000000 16.0000000 16.0000000
MEAN 3.8125000 3.6250000 3.8750000 3.1250000
STDDEV 1.7594980 1.8211718 1.8929694 2.1870833
CORR x1 1.0000000 -.1690411 -.2076656 -.2533673
CORR x2 -.1690411 1.0000000 -.1305323 -.2887231
CORR x3 -.2076656 -.1305323 1.0000000 -.5112620
CORR x4 -.2533673 -.2887231 -.5112620 1.0000000

Number of cases read: 7 Number of cases listed: 7

Notice that this produces a matrix file which is NOT identical to that created by CORRELATIONS ;-((( the RidgeReg Macro (RR) references the N from row 3 (where here they occur in row 1) compute n=x(3,1).
needs to be changed to
compute n=x(*1*,1).

RR assumes the Mean vector is in row 1, where here they are in row 2.
RR assumes the SDs are row 2 where here the end up in Row 3.
compute xmean=x(1,2:(nv+1)).
compute ybar=x(1,1).
compute std=t(x(2,2:(nv+1))).
compute sy=x(2,1).

NEEDS TO BE CHANGED TO
compute xmean=x(*2*,2:(nv+1)).
compute ybar=x(*3*,1).
compute std=t(x(*3*,2:(nv+1))).
compute sy=x(*3*,1).

You will also need to comment out the raw data read in the beginning.

correlations variables=!dep !enter
/missing=listwise
/matrix out(corr).

HTH, David

Maguin, Eugene wrote

> Thanks for all your replies.
> Bruce: In my example, I used the (I,j) values as x(I,j) values to keep
> track of operation. The real matrices are just covariance matrices.
>
> David: It's more easily thought of as a matrix problem, I agree.
>
> Art: I think the answer is No. Matrices are written to the output
> (listing) file in lower triangular but only with three decimal values.
>
> Gene Maguin
>
>
> -----Original Message-----
> From: SPSSX(r) Discussion [mailto:

> SPSSX-L@.UGA

> ] On Behalf Of David Marso
> Sent: Friday, November 15, 2013 6:02 PM
> To:

> SPSSX-L@.UGA

> Subject: Re: matrix data unscrambling problem
>
> I don't see the reluctance to learn MATRIX.
> It is quite powerful and pretty simple once you get a taste.
>
> DATA LIST FREE / V.
> BEGIN DATA
> 11 21 22 31 32
> 33 41 42 43 44
> 51 52 53 54 55
> 61 62 63 64 65
> 66
> END DATA.
> MATRIX.
> GET V / FILE * / VARIABLES V.
> + COMPUTE COV=MAKE(6,6,0).
> + COMPUTE #=1.
> + LOOP #R=1 TO 6.
> + LOOP #C=1 TO #R.
> + COMPUTE COV(#R,#C)=V(#).
> + COMPUTE COV(#C,#R)=V(#).
> + COMPUTE #=#+1.
> + END LOOP.
> + END LOOP.
> PRINT COV.
> END MATRIX.
>
> Run MATRIX procedure:
>
> COV
> 11 21 31 41 51 61
> 21 22 32 42 52 62
> 31 32 33 43 53 63
> 41 42 43 44 54 64
> 51 52 53 54 55 65
> 61 62 63 64 65 66
>
> ------ END MATRIX -----
>
>
>
> DATA LIST FREE / V.
> BEGIN DATA
> 11 21 22 31 32
> 33 41 42 43 44
> 51 52 53 54 55
> 61 62 63 64 65
> 66
> END DATA.
> MATRIX.
> GET V / FILE * / VARIABLES V.
> + COMPUTE COV=MAKE(6,6,0).
> + COMPUTE #=1.
> + LOOP #R=1 TO 6.
> + LOOP #C=1 TO #R.
> + COMPUTE COV(#R,#C)=V(#).
> + COMPUTE #=#+1.
> + END LOOP.
> + END LOOP.
> PRINT COV.
> END MATRIX.
>
> Run MATRIX procedure:
>
> COV
> 11 0 0 0 0 0
> 21 22 0 0 0 0
> 31 32 33 0 0 0
> 41 42 43 44 0 0
> 51 52 53 54 55 0
> 61 62 63 64 65 66
>
> ------ END MATRIX -----
>
>
>
>
> -----
> 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/matrix-data-unscrambling
> -problem-tp5723092p5723099.html Sent from the SPSSX Discussion mailing
> list archive at Nabble.com.
>
> =====================
> 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
>
> =====================
> 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/matrix-data-unscrambling-problem-tp5723092p5723129.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