SPSSX Discussion

Re: How to rotate a base

Classic

List

Threaded

1 message

David Marso

Re: How to rotate a base

Administrator

Rather than going WIDE to LONG, SORT, then LONG to WIDE:
Possibly use AUTORECODE with / GROUP on the product variables.
Use results as INDEX to populate vectors and AGGREGATE.
My prehistoric version of SPSS (11.5) doesn't support /GROUP so this syntax
is just a guess.

AUTORECODE prod_1 prod_2...prod_n
/INTO I_1 I_2...I_n
/GROUP.
VECTOR EVAL (N) /PROD(N).
DO REPEAT I=I_1 TO I_N
/P=Prod_1 Prod_2...Prod_N
/E=EVal_1 Eval_2...Eval_n.
COMPUTE PROD(I)=P.
COMPUTE EVAL(I)=E.
AGGREGATE OUTFILE *
/BREAK ID
/PROD_1 EVAL_1 .... PROD_N EVAL_N
=MAX(PROD_1 EVAL_1 .... PROD_N EVAL_N).

Have no way to test this but it will work if OP does the homework.
IOW: No need to splatter the data, reorder and unsplatter ;-)
This might be more "code" than
VARSTOCASES .
SORT.
CASESTOVARS.
(6 lines vs 3), but is likely to be more efficient .
-Someone should test to verify this over various sizes of files-
Since I have no idea whether these are string or numeric variables, OP
should adjust as necessary (ie RTFM on the VECTOR command for declaring
strings -VECTOR V(N,A##)!!!)

If one were to go totally oldskool -say pre VARSTOCASES AND CASESTOVARS.
This could be done using XSAVE, AUTORECODE, VECTOR and AGGREGATE.
HTH, David

On Sun, 8 Aug 2010 20:48:28 -0400, Richard Ristow <[hidden email]>
wrote:

><html>
><body>
>At 02:36 AM 8/3/2010, Alexandra Chirilov wrote: 
><blockquote type=cite class=cite cite="">
>Have someone a syntax which would
>help me to rotate a database in SPSS? 
>My database show like this: 
> 
>Respondent1   
>Product A      Evaluation for prod
>A      Product B    Evaluation
>for product B   
>Respondent2    Product B     
>Evaluation for prod B      Product
>A    Evaluation for product A </blockquote> 
>Or, shortening the values and reading the data as SPSS variables (and
>correcting Respondent4s 4 and 5, each of whom had the same product
>entered more than once): 
>
>|-----------------------------|---------------------------| 
>|Output
>Created              
>|08-AUG-2010 20:34:59       | 
>|-----------------------------|---------------------------| 
>Respondent  Prod01 Eval01 Prod02 Eval02 
>Respondent1 Prod_A Eval A Prod_B Eval_B 
>Respondent2 Prod_B Eval_B Prod_A Eval_A 
>Respondent3 Prod_C Eval_C Prod_A Eval_A 
>Respondent4 Prod_C Eval_C Prod_X Eval_X 
>Respondent5 Prod_D Eval_D Prod_X Eval_X 
>Respondent6 Prod_A Eval_A Prod_B Eval_B 
>Number of cases read:  6    Number of cases
>listed:  6 
><blockquote type=cite class=cite cite="">
>And I need my database to look
>like: 
>  
>Respondent1    Product A     
>Evaluation for prod A      Product
>B    Evaluation for product B  
>ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½
>(I want to create specific columns for each attribute) 
>Respondent2   
>Product A      Evaluation for prod
>A      Product B    Evaluation
>for product B </blockquote> 
>I'd do this with <tt>VARSTOCASES, SORT, CASESTOVARS.
>I've omitted the columns with "Product A", "Product
>B", etc., because the values in those columns can never vary within
>the data, except between product name and blank. Code is tested. 
>VARSTOCASES 
> /MAKE Prod FROM Prod01 Prod02 
> /MAKE Eval FROM Eval01 Eval02 
> /KEEP =  Respondent 
> /NULL = KEEP. 
>  
>Variables to Cases 
>|-----------------------------|---------------------------| 
>|Output
>Created              
>|08-AUG-2010 20:35:00       | 
>|-----------------------------|---------------------------| 
>Generated Variables 
>|----|------| 
>|Name|Label | 
>|----|------| 
>|Prod|<none>| 
>|Eval|<none>| 
>|----|------| 
>Processing Statistics 
>|-------------|-| 
>|Variables In |5| 
>|Variables Out|3| 
>|-------------|-| 
>  
>SORT CASES BY Respondent Prod. 
>CASESTOVARS 
> /ID      = Respondent 
> /INDEX   = Prod 
> /GROUPBY = INDEX . 
>Cases to Variables 
>|-----------------------------|---------------------------| 
>|Output
>Created              
>|08-AUG-2010 20:35:00       | 
>|-----------------------------|---------------------------| 
>Generated Variables 
>|--------|------|------| 
>|Original|Prod  |Result| 
>|Variable|      |------| 
>|       
>|      |Name  | 
>|--------|------|------| 
>|Eval    |Prod_A|Prod_A| 
>|        |Prod_B|Prod_B| 
>|        |Prod_C|Prod_C| 
>|        |Prod_D|Prod_D| 
>|        |Prod_X|Prod_X| 
>|--------|------|------| 
>Processing Statistics 
>|---------------|---| 
>|Cases In       |12 | 
>|Cases Out      |6  | 
>|---------------|---| 
>|Cases In/Cases |2.0| 
>|Out           
>|   | 
>|---------------|---| 
>|Variables In   |3  | 
>|Variables Out  |6  | 
>|---------------|---| 
>|Index Values   |5  | 
>|---------------|---| 
>  
>LIST. 
>List 
>|-----------------------------|---------------------------| 
>|Output
>Created              
>|08-AUG-2010 20:35:00       | 
>|-----------------------------|---------------------------| 
>Respondent  Prod_A Prod_B Prod_C Prod_D Prod_X 
>Respondent1 Eval A Eval_B 
>Respondent2 Eval_A Eval_B 
>Respondent3 Eval_A        Eval_C 
>Respondent4              
>Eval_C        Eval_X 
>Respondent5                     
>Eval_D Eval_X 
>Respondent6 Eval_A Eval_B 
>Number of cases read:  6    Number of cases
>listed:  6 
></tt>============================= 
>APPENDIX: Test data, and code 
>============================= 
><tt>DATA LIST FIXED/ 
>   Respondent 01-11 (A) 
>   Prod01     14-19
>(A)       
>   Eval01     26-31 (A) 
>   Prod02     38-43
>(A)        
>   Eval02     48-53 (A). 
>*---|---10----|---20----|---30----|---40----|---50----|---60. 
>BEGIN DATA 
>Respondent1  Prod_A      Eval
>A      Prod_B    Eval_B 
>Respondent2  Prod_B     
>Eval_B      Prod_A    Eval_A 
>Respondent3  Prod_C     
>Eval_C      Prod_A    Eval_A 
>Respondent4  Prod_C     
>Eval_C      Prod_X    Eval_X 
>Respondent5  Prod_D     
>Eval_D      Prod_X    Eval_X 
>Respondent6  Prod_A     
>Eval_A      Prod_B    Eval_B 
>END DATA. 
>LIST. 
>VARSTOCASES  
> /MAKE Prod FROM Prod01 Prod02 
> /MAKE Eval FROM Eval01 Eval02 
> /KEEP =  Respondent 
> /NULL = KEEP. 
>LIST. 
>SORT CASES BY Respondent Prod. 
>CASESTOVARS 
> /ID      = Respondent 
> /INDEX   = Prod 
> /GROUPBY = INDEX . 
>LIST.</body>
> 
></html>
>
>====================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

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?"