f



load MATRIX data into normal spss file and get rid of the upper diagonal of the matrix

Hi,

I have a survey sample file with some 260 variables, that is repeated for 2=
1 different subgroups (this could go up to 50). Per subgroup, I create a co=
rrelation matrix over all 260 variables (yes, that can be useful...). I.e.,=
 I use the CORR command to write out a matrix file.

The goal of my intended analysis is to check the pairwise correlations for =
outliers. Outliers are defined in comparison to same variable pairs across =
all subgroups.

Using VARSTOCASES, I can convert the matrix to long format, making all pair=
s of the substantive variables the cases in the new data set. Retaining the=
 subgroup variable, this in principle allows me to do analyses over subgrou=
ps.

(Note that this yields around 1.4 million rows in the long dataset, but onl=
y a few variables.)

Now the problem: CORRELATION gives me a *full* matrix, i.e., variable pairs=
 are duplicated, which of course carries over to the long file: I get rows =
with the variable combination v5-v6 and with v6-v5, which hold the same cor=
relation coefficient. But I obviously only want the n*(n-1)/2 unique variab=
le combinations (per subgroup).

After hours of scratching my head, I have not made any real progress. Can t=
his be done through normal syntax, or must I dive into MATRIX syntax and di=
scard the upper diagonal there, before the reshape of the dataset (however,=
 I am not familiar with MATRIX)?

Thanks for any hints!

Markus

0
Markus
3/26/2015 3:37:22 PM
comp.soft-sys.stat.spss 5679 articles. 0 followers. Post Follow

9 Replies
390 Views

Similar Articles

[PageSpeed] 6

Markus,
MATRIX is actually quite simple once you wrap your brain around it.
It can also get pretty wild once you let it wrap its tentacles around your =
brain.

Here is how you can zap the upper triangle in a few lines of code:
HTH, David
--
MATRIX.
SAVE UNIFORM(1000,100) /OUTFILE * / VARIABLES x001 TO x100.
END MATRIX.
DATASET NAME corrraw.
CORRELATIONS VARIABLES x001 TO x100 /MATRIX OUT (*).
SELECT IF (ROWTYPE_=3D"CORR").
DATASET NAME corrmatrix.

/* above merely simulates some data and creates a CORR matrix to diddle */.
PRESERVE.
SET MXLOOPS=3D10000.
MATRIX.
GET DATA / FILE * / VARIABLES x001 TO x100.
LOOP #=3D1 TO NROW (data)-1.
+  LOOP ##=3D#+1 TO NCOL(data).
+    COMPUTE data(#,##)=3D-999.
+  END LOOP.
END LOOP.
SAVE data / OUTFILE * / VARIABLES x001 TO x100.
END MATRIX.
RESTORE.




On Thursday, March 26, 2015 at 11:37:29 AM UTC-4, Markus wrote:
> Hi,
>=20
> I have a survey sample file with some 260 variables, that is repeated for=
 21 different subgroups (this could go up to 50). Per subgroup, I create a =
correlation matrix over all 260 variables (yes, that can be useful...). I.e=
.., I use the CORR command to write out a matrix file.
>=20
> The goal of my intended analysis is to check the pairwise correlations fo=
r outliers. Outliers are defined in comparison to same variable pairs acros=
s all subgroups.
>=20
> Using VARSTOCASES, I can convert the matrix to long format, making all pa=
irs of the substantive variables the cases in the new data set. Retaining t=
he subgroup variable, this in principle allows me to do analyses over subgr=
oups.
>=20
> (Note that this yields around 1.4 million rows in the long dataset, but o=
nly a few variables.)
>=20
> Now the problem: CORRELATION gives me a *full* matrix, i.e., variable pai=
rs are duplicated, which of course carries over to the long file: I get row=
s with the variable combination v5-v6 and with v6-v5, which hold the same c=
orrelation coefficient. But I obviously only want the n*(n-1)/2 unique vari=
able combinations (per subgroup).
>=20
> After hours of scratching my head, I have not made any real progress. Can=
 this be done through normal syntax, or must I dive into MATRIX syntax and =
discard the upper diagonal there, before the reshape of the dataset (howeve=
r, I am not familiar with MATRIX)?
>=20
> Thanks for any hints!
>=20
> Markus

0
David
3/26/2015 6:53:27 PM
Am Freitag, 27. M=E4rz 2015 01:26:03 UTC+1 schrieb Markus:
> Ah, I knew the looping would be simple in MATRIX, but I thought I could s=
omehow emulate that with 'standard' syntax. However, I totally failed to cr=
eate something that serves the function of the row and col indices...
>=20
> Thank you so much, anyway, I'll try your code tomorrow!
>=20
> Best,
> Markus

PS: and no, I wouldn't have managed myself to write that MATRIX code in les=
s than a very long day. Thanks again!
0
Markus
3/27/2015 12:28:35 AM
Ah, I knew the looping would be simple in MATRIX, but I thought I could somehow emulate that with 'standard' syntax. However, I totally failed to create something that serves the function of the row and col indices...

Thank you so much, anyway, I'll try your code tomorrow!

Best,
Markus
0
Markus
3/27/2015 1:01:01 AM
On Thu, 26 Mar 2015 08:37:22 -0700 (PDT), Markus <mquandt20@gmail.com>
wrote:

>
>The goal of my intended analysis is to check the pairwise correlations for outliers. Outliers are defined in comparison to same variable pairs across all subgroups.

Outliers among correlations?  I have never searched for those, 
but naturally I have run across a few.  Usually, the differences
were properly blamed on artifacts, not on interesting 
substance of the data. 

The ones that I saw in other people's data were sometimes
owing to invalid scores being included, either using the 
"-999" missing as real data, or some mis-entered score. 

Real outliers also cause odd correlations; sometimes that
is a sign that the data should have been transformed
(take a log, reciprocal, or whatever).  Sets of skewed data
will not be homogeneous for variance.   Further, if scores are
so skewed that "a few" cases account for 50 or 80% of the
total variance, then the r may act like has 2 or 3 d.f.; seeing
a correlation of 0.70  vs  0.0  may be a trivial difference.

The harder artifact to spot is one a particular sample happens
to have very small variance on one variable: That is why
a "test for equal regression coefficient"  is often more 
meaningful than a test for equal correlations. 

Before I bothered scanning hundreds of sets of correlations,
and after I checked for min/max to check for invalid data (and
generally homogeneous subgroups), I would compare the 
variances between subgroups to find outliers by this simpler 
criterion, variance....  Differences in variance will subvert most
of the meaning that we expect when speaking of similar
"correlation".  

-- 
Rich Ulrich 
0
Rich
3/27/2015 1:01:01 AM
Am Freitag, 27. M=E4rz 2015 06:53:29 UTC+1 schrieb Rich Ulrich:
> On Thu, 26 Mar 2015 08:37:22 -0700 (PDT), Markus
> wrote:
>=20
> >
> >The goal of my intended analysis is to check the pairwise correlations f=
or outliers. Outliers are defined in comparison to same variable pairs acro=
ss all subgroups.
>=20
> Outliers among correlations?  I have never searched for those,=20
> but naturally I have run across a few.  Usually, the differences
> were properly blamed on artifacts, not on interesting=20
> substance of the data.=20
>=20
> The ones that I saw in other people's data were sometimes
> owing to invalid scores being included, either using the=20
> "-999" missing as real data, or some mis-entered score.=20
>=20
> Real outliers also cause odd correlations; sometimes that
> is a sign that the data should have been transformed
> (take a log, reciprocal, or whatever).  Sets of skewed data
> will not be homogeneous for variance.   Further, if scores are
> so skewed that "a few" cases account for 50 or 80% of the
> total variance, then the r may act like has 2 or 3 d.f.; seeing
> a correlation of 0.70  vs  0.0  may be a trivial difference.
>=20
> The harder artifact to spot is one a particular sample happens
> to have very small variance on one variable: That is why
> a "test for equal regression coefficient"  is often more=20
> meaningful than a test for equal correlations.=20
>=20
> Before I bothered scanning hundreds of sets of correlations,
> and after I checked for min/max to check for invalid data (and
> generally homogeneous subgroups), I would compare the=20
> variances between subgroups to find outliers by this simpler=20
> criterion, variance....  Differences in variance will subvert most
> of the meaning that we expect when speaking of similar
> "correlation". =20
>=20
> --=20
> Rich Ulrich

Rich,

Thanks for the comment. Yes, it IS exactly possible artificats, among them =
prominently reversed codings for subsample variables, that I am looking for=
, in a brute force approach. More substantive error (or explanation) checki=
ng begins after we identified possibly suspicious correlations.

Markus
0
Markus
3/27/2015 8:48:00 AM
Am Donnerstag, 26. M=E4rz 2015 19:53:31 UTC+1 schrieb David Marso:
> Markus,
> MATRIX is actually quite simple once you wrap your brain around it.
> It can also get pretty wild once you let it wrap its tentacles around you=
r brain.
>=20
> Here is how you can zap the upper triangle in a few lines of code:
> HTH, David
> --
> MATRIX.
> SAVE UNIFORM(1000,100) /OUTFILE * / VARIABLES x001 TO x100.
> END MATRIX.
> DATASET NAME corrraw.
> CORRELATIONS VARIABLES x001 TO x100 /MATRIX OUT (*).
> SELECT IF (ROWTYPE_=3D"CORR").
> DATASET NAME corrmatrix.
>=20
> /* above merely simulates some data and creates a CORR matrix to diddle *=
/.
> PRESERVE.
> SET MXLOOPS=3D10000.
> MATRIX.
> GET DATA / FILE * / VARIABLES x001 TO x100.
> LOOP #=3D1 TO NROW (data)-1.
> +  LOOP ##=3D#+1 TO NCOL(data).
> +    COMPUTE data(#,##)=3D-999.
> +  END LOOP.
> END LOOP.
> SAVE data / OUTFILE * / VARIABLES x001 TO x100.
> END MATRIX.
> RESTORE.
>=20
>=20
>=20
>=20
> On Thursday, March 26, 2015 at 11:37:29 AM UTC-4, Markus wrote:
> > Hi,
> >=20
> > I have a survey sample file with some 260 variables, that is repeated f=
or 21 different subgroups (this could go up to 50). Per subgroup, I create =
a correlation matrix over all 260 variables (yes, that can be useful...). I=
..e., I use the CORR command to write out a matrix file.
> >=20
> > The goal of my intended analysis is to check the pairwise correlations =
for outliers. Outliers are defined in comparison to same variable pairs acr=
oss all subgroups.
> >=20
> > Using VARSTOCASES, I can convert the matrix to long format, making all =
pairs of the substantive variables the cases in the new data set. Retaining=
 the subgroup variable, this in principle allows me to do analyses over sub=
groups.
> >=20
> > (Note that this yields around 1.4 million rows in the long dataset, but=
 only a few variables.)
> >=20
> > Now the problem: CORRELATION gives me a *full* matrix, i.e., variable p=
airs are duplicated, which of course carries over to the long file: I get r=
ows with the variable combination v5-v6 and with v6-v5, which hold the same=
 correlation coefficient. But I obviously only want the n*(n-1)/2 unique va=
riable combinations (per subgroup).
> >=20
> > After hours of scratching my head, I have not made any real progress. C=
an this be done through normal syntax, or must I dive into MATRIX syntax an=
d discard the upper diagonal there, before the reshape of the dataset (howe=
ver, I am not familiar with MATRIX)?
> >=20
> > Thanks for any hints!
> >=20
> > Markus


Hi,

having now tried it, I cannot get it to run. It even starts with the first =
block, which gives me this:

>Error # 4096.  Command name: MATRIX=20
>No permanent dictionary has been defined.  There are no variables defined =
in=20
>the above transformations that a procedure may access.=20
>Execution of this command stops.=20
=20
Rest of MATRIX input skipped.

If I try to run the second block on my data (after exchanging variable name=
s and removing the indentation, which SPSS 22.0.0.1 win 64 refuses to execu=
te), I don't get an error, but virtually nothing happens to my data.

Lost,
Markus
0
Markus
3/27/2015 2:04:46 PM
On 26/03/2015 11:37 AM, Markus wrote:
> Hi,
>
> I have a survey sample file with some 260 variables, that is repeated for 21 different subgroups (this could go up to 50). Per subgroup, I create a correlation matrix over all 260 variables (yes, that can be useful...). I.e., I use the CORR command to write out a matrix file.
>
> The goal of my intended analysis is to check the pairwise correlations for outliers. Outliers are defined in comparison to same variable pairs across all subgroups.
>
> Using VARSTOCASES, I can convert the matrix to long format, making all pairs of the substantive variables the cases in the new data set. Retaining the subgroup variable, this in principle allows me to do analyses over subgroups.
>
> (Note that this yields around 1.4 million rows in the long dataset, but only a few variables.)
>
> Now the problem: CORRELATION gives me a *full* matrix, i.e., variable pairs are duplicated, which of course carries over to the long file: I get rows with the variable combination v5-v6 and with v6-v5, which hold the same correlation coefficient. But I obviously only want the n*(n-1)/2 unique variable combinations (per subgroup).
>
> After hours of scratching my head, I have not made any real progress. Can this be done through normal syntax, or must I dive into MATRIX syntax and discard the upper diagonal there, before the reshape of the dataset (however, I am not familiar with MATRIX)?
>
> Thanks for any hints!
>
> Markus
>

Instead of getting rid of the upper diagonal, you could find and remove 
duplicate correlations (in the GUI:  DATA > Identify Duplicate Cases). 
The assumption here is that no correlations would be identical to 
however many decimals SPSS carries except the ones that are structurally 
identical (e.g., X-Y and Y-X correlation).

HTH.


-- 
Bruce Weaver
bweaver@lakeheadu.ca
http://sites.google.com/a/lakeheadu.ca/bweaver/Home
"When all else fails, RTFM."
0
Bruce
3/27/2015 3:47:36 PM
On Fri, 27 Mar 2015 01:48:00 -0700 (PDT), Markus <mquandt20@gmail.com>
wrote:

>Am Freitag, 27. März 2015 06:53:29 UTC+1 schrieb Rich Ulrich:
>> On Thu, 26 Mar 2015 08:37:22 -0700 (PDT), Markus
>> wrote:
>> 
>> >
>> >The goal of my intended analysis is to check the pairwise correlations for outliers. Outliers are defined in comparison to same variable pairs across all subgroups.
>> 
>> Outliers among correlations?  I have never searched for those, 
>> but naturally I have run across a few.  Usually, the differences
>> were properly blamed on artifacts, not on interesting 
>> substance of the data. 
>> 
>> The ones that I saw in other people's data were sometimes
>> owing to invalid scores being included, either using the 
>> "-999" missing as real data, or some mis-entered score. 
>> 
>> Real outliers also cause odd correlations; sometimes that
>> is a sign that the data should have been transformed
>> (take a log, reciprocal, or whatever).  Sets of skewed data
>> will not be homogeneous for variance.   Further, if scores are
>> so skewed that "a few" cases account for 50 or 80% of the
>> total variance, then the r may act like has 2 or 3 d.f.; seeing
>> a correlation of 0.70  vs  0.0  may be a trivial difference.
>> 
>> The harder artifact to spot is one a particular sample happens
>> to have very small variance on one variable: That is why
>> a "test for equal regression coefficient"  is often more 
>> meaningful than a test for equal correlations. 
>> 
>> Before I bothered scanning hundreds of sets of correlations,
>> and after I checked for min/max to check for invalid data (and
>> generally homogeneous subgroups), I would compare the 
>> variances between subgroups to find outliers by this simpler 
>> criterion, variance....  Differences in variance will subvert most
>> of the meaning that we expect when speaking of similar
>> "correlation".  
>> 

>
>Rich,
>
>Thanks for the comment. Yes, it IS exactly possible artificats, among them prominently reversed codings for subsample variables, that I am looking for, in a brute force approach. More substantive error (or explanation) checking begins after we identified possibly suspicious correlations.
>
>Markus

Okay, reversed codings isn't a problem that I have ever faced
on any large scale.  

I will just add:  When I have had hundreds of variables, they
have always included a bunch of rating scales.  If the problem
has  whole scales whose items might be reversed, I would 
prefer to work from the scored-up composite than from the
original items, since the composite will have a more reliable
score.  And there will be fewer of them to look at. 

-- 
Rich Ulrich 

0
Rich
3/27/2015 8:50:20 PM
Important to have the MATRIX system file active.

This runs without any errors.
MATRIX.
SAVE UNIFORM(1000,100) /OUTFILE * / VARIABLES x001 TO x100.
END MATRIX.
DATASET NAME corrraw.
CORRELATIONS VARIABLES x001 TO x100 /MATRIX OUT (*).
SELECT IF (ROWTYPE_=3D"CORR").
EXECUTE.
DATASET NAME corrmatrix.

/* above merely simulates some data and creates a CORR matrix to diddle */.
PRESERVE.
SET MXLOOPS=3D10000.
DELETE VARIABLES ROWTYPE_ VARNAME_ .
MATRIX.
GET DATA / FILE * / VARIABLES ALL /NAMES=3Dnamevec.
LOOP #=3D1 TO NROW (data)-1.
LOOP ##=3D#+1 TO NCOL(data).
COMPUTE data(#,##)=3D-999.
END LOOP.
END LOOP.
SAVE data / OUTFILE * / NAMES =3Dnamevec.
END MATRIX.
RESTORE.


On Friday, March 27, 2015 at 10:04:51 AM UTC-4, Markus wrote:
> Am Donnerstag, 26. M=E4rz 2015 19:53:31 UTC+1 schrieb David Marso:
> > Markus,
> > MATRIX is actually quite simple once you wrap your brain around it.
> > It can also get pretty wild once you let it wrap its tentacles around y=
our brain.
> >=20
> > Here is how you can zap the upper triangle in a few lines of code:
> > HTH, David
> > --
> > MATRIX.
> > SAVE UNIFORM(1000,100) /OUTFILE * / VARIABLES x001 TO x100.
> > END MATRIX.
> > DATASET NAME corrraw.
> > CORRELATIONS VARIABLES x001 TO x100 /MATRIX OUT (*).
> > SELECT IF (ROWTYPE_=3D"CORR").
> > DATASET NAME corrmatrix.
> >=20
> > /* above merely simulates some data and creates a CORR matrix to diddle=
 */.
> > PRESERVE.
> > SET MXLOOPS=3D10000.
> > MATRIX.
> > GET DATA / FILE * / VARIABLES x001 TO x100.
> > LOOP #=3D1 TO NROW (data)-1.
> > +  LOOP ##=3D#+1 TO NCOL(data).
> > +    COMPUTE data(#,##)=3D-999.
> > +  END LOOP.
> > END LOOP.
> > SAVE data / OUTFILE * / VARIABLES x001 TO x100.
> > END MATRIX.
> > RESTORE.
> >=20
> >=20
> >=20
> >=20
> > On Thursday, March 26, 2015 at 11:37:29 AM UTC-4, Markus wrote:
> > > Hi,
> > >=20
> > > I have a survey sample file with some 260 variables, that is repeated=
 for 21 different subgroups (this could go up to 50). Per subgroup, I creat=
e a correlation matrix over all 260 variables (yes, that can be useful...).=
 I.e., I use the CORR command to write out a matrix file.
> > >=20
> > > The goal of my intended analysis is to check the pairwise correlation=
s for outliers. Outliers are defined in comparison to same variable pairs a=
cross all subgroups.
> > >=20
> > > Using VARSTOCASES, I can convert the matrix to long format, making al=
l pairs of the substantive variables the cases in the new data set. Retaini=
ng the subgroup variable, this in principle allows me to do analyses over s=
ubgroups.
> > >=20
> > > (Note that this yields around 1.4 million rows in the long dataset, b=
ut only a few variables.)
> > >=20
> > > Now the problem: CORRELATION gives me a *full* matrix, i.e., variable=
 pairs are duplicated, which of course carries over to the long file: I get=
 rows with the variable combination v5-v6 and with v6-v5, which hold the sa=
me correlation coefficient. But I obviously only want the n*(n-1)/2 unique =
variable combinations (per subgroup).
> > >=20
> > > After hours of scratching my head, I have not made any real progress.=
 Can this be done through normal syntax, or must I dive into MATRIX syntax =
and discard the upper diagonal there, before the reshape of the dataset (ho=
wever, I am not familiar with MATRIX)?
> > >=20
> > > Thanks for any hints!
> > >=20
> > > Markus
>=20
>=20
> Hi,
>=20
> having now tried it, I cannot get it to run. It even starts with the firs=
t block, which gives me this:
>=20
> >Error # 4096.  Command name: MATRIX=20
> >No permanent dictionary has been defined.  There are no variables define=
d in=20
> >the above transformations that a procedure may access.=20
> >Execution of this command stops.=20
> =20
> Rest of MATRIX input skipped.
>=20
> If I try to run the second block on my data (after exchanging variable na=
mes and removing the indentation, which SPSS 22.0.0.1 win 64 refuses to exe=
cute), I don't get an error, but virtually nothing happens to my data.
>=20
> Lost,
> Markus

I assumes the MATRIX system file is the active file.
0
David
3/28/2015 1:44:28 AM
Reply:

Similar Artilces:

matrix in a matrix
hallo there. I want to put a matrix that is 128X128 in a matrix that is 512X512 ( ones(512) ). what I want to do is to have the white image and above to put the second image ( a letter, A for example). I cant multiply cause they dont have the same dimensions. any ideas?? thanks in advance "th " <THOMITSU@YAHOO.COM> wrote in message <hsjabs$bbc$1@fred.mathworks.com>... > hallo there. > > I want to put a matrix that is 128X128 in a matrix that is 512X512 ( ones(512) ). > > what I want to do is to have the white image and above ...

Matrix in a matrix?
Hi to all, I've encountered a problem using Matlab. As I'm not very familiar with matlab, there are functions which I have not used before. What I would like to know, if there is a way to express a matrix within a matrix. For example, R= [ r1 r2 r3 ... rn] where rn=[xn yn tn] and xn, yn, tn are column of numbers. I used the following code below to express the numbers in rn as coloumns for easy referencing. xn=r(:,1); %represents the columns in the matrix yn=r(:,2); tn=r(:,3); My objective is to be able to retrieve information in the matrix easily and compare their values. Meani...

matrix in matrix (?)
Hi, I need help! I want to construct a matrix 10x10 where every element is a matrix ?x2. After, I need to extract the ?x2 matrix by the 10x10. how to make that? mirus wrote: > Hi, I need help! > > I want to construct a matrix 10x10 where every element is a matrix > ?x2. > > After, I need to extract the ?x2 matrix by the 10x10. > > how to make that? have a look at cell arrays. they provide the possibility to store arbitrary data in an array like fashion. michael Michael Wild wrote: > mirus wrote: >> Hi, I need help! >> >> I want to constru...

Matrix of matrix
Hi everbody. First of all, I apologize for my English, so please be indulgent with me if my sentences are not grammatically correct. I have some trouble with MATLAB. I have some data in txt files that I catch like that : MFS93A1 = textread('Z:/Fichiers/MFS93.A1.txt','','headerlines',1); MFS93B1 = textread('Z:/Fichiers/MFS93.B1.txt','','headerlines',1); MFM77A1 = textread('Z:/Fichiers/MFM77.A1.txt','','headerlines',1); MFM77C1 = textread('Z:/Fichiers/MFM77.C1.txt','','headerlines',1); Then, in or...

SPSS data file to SAS data file: only partial variable labels
Dear fellow SAS users: I am trying to convert an SPSS data file (.sav, not .por) to SAS, using SPSS version 15 for Windows. The data file is large, but not huge (176 cases, 963 variables). I have SAS 9.1.3 on Windows XP. I used the following commands, and got the Warning message that "excess labels will be omitted." (See below) SPSS successfully copied over the variable labels for the first 2/3 of the variables, but then just stopped. The wording of the message ("cumulative length of the variable labels exceeds the limitations of the target file type") suggests that there...

Empty matrix and Diagonal matrix
Hi everyone I want to creat a large diagonal matrix.The main diagonal elements of the diagonal matrix is a 3x3 matrix. For example I have a matrix A(3x3),now I should obtain the diagnoal matrix B(300x300),can anybody tell me how to do it? Deeply appreciate your help! Thank you in advance! Nandy Erik wrote: > > > Hi everyone > I want to creat a large diagonal matrix.The main diagonal > elements > of the diagonal matrix is a 3x3 matrix. > For example I have a matrix A(3x3),now I should obtain the diagnoal > matrix B(300x300), >Others I want to know how to create an...

1) How do I create a matrix of both text and numbers, and 2) How do I get my GUI to add data to the matrix
I have a library of data in the form of a matrix, and in my GUI, I want the user to be able to pick a name from a listbox, and all the info attached to that name can be brought up in the GUI. This bit is not so much the problem because, I have an idea in my head of how to do this (although any suggestions would be helpful). The problem is that I want the user to be able to add more entries to my library of data and I'm not sure how to do this because I don't know how to get my matrix to store names. I think I could work something out if I could get an array to store both text and num...

Pls help
Hi, I have a text file that has 544 entry information like this: entry-1: A/L R/K N/M D/F C/P Q/S E/T G/W 4.35 4.38 4.75 4.76 4.65 4.37 4.29 3.97 4.17 4.36 4.52 4.66 4.44 4.50 4.35 4.70 Here, A has 4.35 value, L has 4.17 value, R has 4.38, K has 4.36 an so on. Now, I want to convert this text file into the MAT file(double array or matrix). In this MAT file A,R,N,D,C,Q,E,G,L,K,M,F,P,S,T,W should be the rows and the related values should be the columns. It's easy for one entry but my text file composed of 544 entries like...

Pls help
Hi, I have a text file that has 544 entry information like this: entry-1: A/L R/K N/M D/F C/P Q/S E/T G/W H/Y I/V 4.35 4.38 4.75 4.76 4.65 4.37 4.29 3.97 4.63 3.95 4.17 4.36 4.52 4.66 4.44 4.50 4.35 4.70 4.60 3.95 Here, A has 4.35 value, L has 4.17 value, R has 4.38, K has 4.36 an so on. Now, I want to convert this text file into the MAT file(double array or matrix). In this MAT file A,R,N,D,C,Q,E,G,H,I,L,K,M,F,P,S,T,W,Y,V should be the rows and the related values should be the columns. It's easy for one entry b...

ANN: Neural Data Visualisation Software now loads SPSS & SAS files
Dear all, Tiberius is a neural network modelling tool and also a very useful data visualisation tool. There are currently over 7,000 users worldwide who have nothing but praise for the software, and its very reasonable $15 price tag (the limited version is free). Tiberius can now load SPSS files directly if the SPSS data access pack has been installed. It can also load SAS files. Please download Tiberius and give it a try. Tiberius can be found at... http://www.philbrierley.com ...

MATRIX : READING UPPER TRIANGULAR MATRIX
Hello Friends, Can you suggest me any way using which I can read the upper triangular matrix. meaning if the matrix is A= [1 2 3 4; 2 3 4 5; 6 7 6 8; 1 2 3 4 ] then I wish to have output as [ 2 3 4 4 5 8 ] meaning without reading the diagonal and the lower triangular matrix. Looking forward to your suggestions Nishant "Nishant" <nishantxl@rediffmail.com> wrote in message news:1127441169.021440.234280@z14g2000cwz.googlegroups.com... > Hello Friends, > Can you suggest me any way using which I can read the > upper triang...

Convert an SPSS sav file to SPSS portable files so I can read in SAS
I was wondering if someone would be willing to convert a couple of very small (1200 records) SPSS *sav files to a portable file format so that I can build a SAS dataset from it. On Jan 13, 8:34=A0am, Takeadoe <mtonkov...@msn.com> wrote: > I was wondering if someone would be willing to convert a couple of > very small (1200 records) SPSS *sav files to a portable file format so > that I can build a SAS dataset from it. Assuming that you have SPSS, you can use SaveAs to create the SAS files directly. No need to go through the portable file format. Takeadoe <mtonkov...

Creating a J*J Diagonal Matrix with Jth Diagonal element is also a matrix
Hello, I am a new beginner of Matlab and I want to create a Diagnoal Matrix with the main diagnoal as (A1,A2,A3,....,Aj) where Aj is also a matrix which is defined as (a11,a12,a13;a21,a22,a23;a31,a32,a33). How can I define Aj as a variable in the m file then create the diagnoal matrix through Aj? Could someone help on this please? Many Thanks in Advance! "Jie " <jz286@cam.ac.uk> wrote in message <h6selu$9ng$1@fred.mathworks.com>... > Hello, > > I am a new beginner of Matlab and I want to create a Diagnoal Matrix with the main diagnoal as (A1,A2,A3,....,Aj) w...

Bitmap data extraction (matrix of matrix)
Hello, This is most likely a simple question for those who use Mathematica on a semi-regular basis, but ... I have a 512x512 matrix composed of 1x3 matrix elements, and want to create a 512x512 matrix composed of a single value in the original 1x3 elements. That doesn't look very well defined, so please let me explain further in the hopes of clarifying what I am after. It's a bitmap image file (512x512) and I want to extract a part of the "raw" data for further processing. The raw data in the original file consists of values for the red, green and blue (RGB)...

Web resources about - load MATRIX data into normal spss file and get rid of the upper diagonal of the matrix - comp.soft-sys.stat.spss

Diagonal - Wikipedia, the free encyclopedia
A diagonal is a line joining two nonconsecutive vertices of a polygon or polyhedron . Informally, any sloping line is called diagonal. The word ...

Number Join + : Join every single number. Horizontal, vertical and diagonal on the App Store on iTunes ...
Get Number Join + : Join every single number. Horizontal, vertical and diagonal on the App Store. See screenshots and ratings, and read customer ...

Police investigate four deliberately lit fires at Rider St, Sandery Ave and Diagonal Rd, Seacombe Gardens ...
POLICE need information to find who deliberately lit four small rubbish or grass fires at Seacombe Gardens early on Tuesday.

WSJ again claims Apple working on larger-screened iPhones up to 6" diagonal [u]
Apple is reportedly testing a variety of next-generation iPhones with larger screen sizes, some up to six inches diagonally, though the devices ...

SteamVR instructions show wireless controllers, 15’ diagonal playspace
New headset has less weight, fewer wires, and Wii-like wrist straps on the controls.

2015 BMW X4 slides in on the diagonal starting at $45,625 in the US
Filed under: New York Auto Show , Crossover , BMW , Luxury If BMW has shown us anything, it's a knack for creating niches and expanding its ...

UAB Blazers Make Tremendous Diagonal Pass to Set Up Big-Time 3-Pointer - Bleacher Report
... aren't giving up yet as they battle it out with the UCLA Bruins in their third-round matchup. Chris Cokley sets up Denzell Watts with the diagonal ...

Witness: Plane did "diagonal cartwheel"
Eyewitness Anthony Castrani describes how Asiana flight 214 performed a "side motion diagonal cartwheel" as it crashed while landing at San Francisco ...

Wind-toppled semitrailer to remain alongside Boulder-bound Diagonal until Tuesday
High winds toppled a semitrailer's load on the Boulder-bound Diagonal Highway near Niwot this afternoon, forcing traffic to divert through the ...

City Rejects Plea to Keep Diagonal Runways to Reduce O'Hare Noise
... are used at night to reduce the noise over some neighborhoods. O'HARE — O'Hare Airport officials Friday rejected pleas to keep two diagonal ...

Resources last updated: 1/25/2016 8:26:28 PM