Using Var Names to Conditionally Build String Values Case By Case...

[Hal 9000]

Hello Everybody,


I have a  program that builds an error report. IIn addition to a bunch of
descriptive variables (ID, Name, Student Type, Major) it is comprised of a
series of Boolean variables with descriptive variable names. For example:



Missing_Stu_Level         Missing_Stu_Type         Missing_Stu_Status

----------------------------------------------------------------------------------------------------------

1                                  0                                  1

1                                  1                                  0

0                                  0                                  0

1                                  1                                  1





This program breaks out repeated subgroups from the dataset based on values
on a series of error codes and appends them to an Excel file. The final
Excel report looks like this:



Data Cleaning Report

ID         Name                Student_Type
Student_Status              Major                Timecode          Advisor

---------------------------------------------------------------------------------------------------------------------------------------------------------------------------

(Missing_Stu_Level)

1234     Joe Doe            DP
CN                                BIO
FT                    Smith

2345     Jane Cane         PA
UF                                PSY                 HT
White

(Missing_Stu_Type)

3456     Will Do              DP
UT                                MUS                 FT
Smith

4567     Ima Tired           DP
UT                                ART                  HT
White

Etc………



As you can see, records are grouped by error code, so if they have multiple
error codes they might appear several times at different places in this
file.



I would like to compute a final variable that indicates which error codes a
person has a value of 1 on. This is so the individual who is doing the data
cleaning can see what other missing data needs to be fixed for any specific
individual. This way they can fix all the errors for each person at once
rather than have to call up the record more than one time due to duplicate
appearances for more than one type of error. I want to create a single
string variable, perhaps 200 characters wide, that will list all the
relevant error variables.

For example:



(The Dataset From Which The Report Is Created)

ID         Missing_Stu_Level   Missing_Stu_Type   Missing_Stu_Status
Multi_Errors

---------------------------------------------------------------------------------------------------------------------------------------------------------------------

####     1                            0
1                         Missing_Stu_Level, Missing_Stu_Status

####     1                            1
0                         Missing_Stu_Level, Missing_Stu_Type

####     0                            0
0

####     1                            1
1                         Missing_Stu_Level, Missing_Stu_Type,
Missing_Stu_Status



Any idea are welcome, including better ways of organizing such a report. I'm
out of ideas for the moment. Thanks!



Gary Moser, Research Associate

Institutional Research

Dominican University

Tel:  [phone]

Fax: [phone]

E-Mail: [hidden email]

[Richard Ristow]

At 03:06 AM 8/20/2006, Hal 9000 wrote:

>I have a  program that builds an error report.
>II addition to a bunch of descriptive variables
>(ID, Name, Student Type, Major) it is comprised
>of a series of Boolean variables with
>descriptive variable names. For example:
>
>Missing_Stu_Level  Missing_Stu_Type Missing_Stu_Status
>-----------------------------------------------------
>1                  0                1
>1                  1                0
>0                  0                0
>1                  1                1
>This program breaks out repeated subgroups from
>the dataset based on values on a series of error
>codes and appends them to an Excel file. The
>final Excel report looks like this:
>
>
>Data Cleaning Report
>
>ID   Name       S_type S_Stat Major Timecode Advisor (Msg_Stu_Lvl)
>------------------------------------------------------------
>1234 Joe Doe    DP            BIO   FT       Smith
>2345 Jane Cane  PA      UF    PSY   HT       White
>
>(Missing_Stu_Type)
>3456 Will Do    DP      UT    MUS   FT       Smith
>4567 Ima Tired  DP      UT    ART   HT       White
>Etc………
>
>As you can see, records are grouped by error
>code, so if they have multiple error codes they
>might appear several times at different places in this file.
>
>I would like to compute a final variable that
>indicates which error codes a person has a value
>of 1 on. This is [one] can fix all the errors
>for each person at once. I want to create a
>single string variable, perhaps 200 characters
>wide, that will list all the relevant error [conditions].

I have, by the way, shortened your variable names
and data lines. For reference, the software that
distributes postings breaks all lines at 72
characters; posted data and code lines are much
more readable if kept within that limit.

What you want ought to be very easy. Try this (not tested):

STRING ALL_ERRS(A200).

DO REPEAT
   ERR_VBL = Missing_Stu_Level  Missing_Stu_Type Missing_Stu_Status
  /ERR_MSG = 'Msg level'        'Msg Type'       'Msg Status'.
.  DO IF    ERR_VBL EQ 1.
.     DO IF   ALL_ERRS EQ ' '.
.        COMPUTE ALL_ERRS = LTRIM(ERR_MSG).
.     ELSE.
.        COMPUTE ALL_ERRS = CONCAT(RTRIM(ALL_ERRS),', ',
                                    LTRIM(ERR_MSG)).

.     END.
.  END IF.
END REPEAT.

SELECT IF ALL_ERRS NE ' '.

I'd keep the error messages short, shorter than
your variable names, so the error-message string won't get unwieldy.

Good luck,
Richard

>For example:
>
>
>
>(The Dataset From Which The Report Is Created)
>
>ID         Missing_Stu_Level   Missing_Stu_Type   Missing_Stu_Status
>Multi_Errors
>
>----------------------------------------------------------------------------------------------------------------------------------------------------------------------
>
>####     1                            0
>1                         Missing_Stu_Level, Missing_Stu_Status
>
>####     1                            1
>0                         Missing_Stu_Level, Missing_Stu_Type
>
>####     0                            0
>0
>
>####     1                            1
>1                         Missing_Stu_Level, Missing_Stu_Type,
>Missing_Stu_Status
>
>
>
>Any idea are welcome, including better ways of
>organizing such a report. I'm
>out of ideas for the moment. Thanks!
>
>
>
>Gary Moser, Research Associate
>
>Institutional Research
>
>Dominican University
>
>Tel:  [phone]
>
>Fax: [phone]
>
>E-Mail: [hidden email]

[Hal 9000]

>
>  Thanks to everybody--all the input I receive jogged my mind, and I came
> up with something that does what I wanted (listed below). I should have
> mentioned that I prefered to leave the values in the error variables as 1's
> and 0's because a macro that comes later is set up to expect those values,
> therefore the most efficient solution for me leaves the data as it is.
> *************************************************************************.
>
*Key: TRANS = Temporary Transfer Variable
         T_ALL_ERRS = Temporary All-Error Variable
         ALL_ERRS = Final Variable Listing All Errors.



> string TRANS (a40) ALL_ERRS (a200) T_ALL_ERRS (a200).
> exe.
>

> define OLAP (OL = !charend ('/'))
> !do !S !in (!OL)
> compute T_ALL_ERRS = ALL_ERRS.
> exe.
> do if (!S = 1).
> compute TRANS = !quote(!S).
> compute ALL_ERRS = concat(rtrim(TRANS), ' / ', rtrim(T_ALL_ERRS)).
> end if.
> exe.
> !doend
> !enddefine.
>
> OLAP OL = Missing_Stu_Level Missing_Ethnicity Missing_SSN Missing_Religion
> Missing_Citizenship Missing_Handicap Missing_Phone1 Missing_City
> Missing_State Missing_Email Missing_Visa_Code Missing_Residence
> Missing_Stu_Status
> Missing_Timecode Missing_Maj1 Citizen_Eth_Conflict Citizen_Visa_Conflict
> /.
>