About data_null__

data_null__ · ‎08-27-2009

> Humbly, you still have not explained why it is > important to know or even care how SAS stores the > information internally, even if you had the > capability or control. However, as mentioned, the > RETAIN statement, though I don't recommend it, will > give you to establish a "default order" (not to be > confused with any SAS-managed internal sequence) for > the variables/columns in your SAS dataset/table. I think the OP means VARNUM order. We know that SAS doesn't necessary store the columns in VARNUM order but that is unimportant for this discussion. VARNUM order is not important most of the time but there are times when clients want variables in a certain order. When that is the case I usually create the variable in the proper order. However there are times when it is necessary to move the columns around, and for that the RETAIN statement is the most concise method. I know that some SAS users have "retain-a-phobia". Perhaps the following example will help. [pre] proc contents data=sashelp.shoes order=varnum; run; data newShoes; stop; retain Inventory Region Product Subsidiary Stores Sales Returns ; set sashelp.shoes; run; data newShoes; set newShoes sashelp.shoes; *** Perhaps more code; run; proc contents data=newShoes order=varnum; run; [/pre]

data_null__ · ‎08-26-2009

> I run following code: > data admit; > input ID $ 1-4 Name $ 6-19 Sex $ 21 Age 23-24 > Date 26-27 Height 29-30 Weight 32-34 > ActLevel $ 36-39 Fee 41-46; > format fee 6.2; > run; > > I guess 'ID $1-4',it will produce a parameter,its > width is 4,but I don't understand 'Sex $21' width is > 1,why? And 'Weight 32-34' width is 8,why? > these are column number ranges and in SAS without other statements to define the variable attributes the column numbers serve that purpose. $1-4 character length 4 $21 same as $21-21 character length 1 32-34 NO $ sign numeric. All numerics default to lenght 8. > > Where is 5,20,22,25,28,31,35,40? Why omit them? Those columns are skipped. No mistery. You need to read about COLUMN INPUT. That will answer most of your questions. This INPUT statement syntax was developed with the very first SAS version from the late sixties. If may seem odd to some.

data_null__ · ‎08-25-2009

> but I have never heard of SAS Variable List before. I'll bet you have used Numbered Range Lists (aka enumerated lists) and possibly the Name Range List, but you may not have seen this section of the documentation. http://tinyurl.com/mtumkg http://support.sas.com/documentation/cdl/en/lrcon/61722/HTML/default/a000695105.htm As I mentioned using these list can be a powerful SAS programming tool.

data_null__ · ‎08-24-2009

> I wanted to have an idea if I can translate the array > as a whole to optimize time runing. It appears that I > can't do this !! 😞 ! Are you sure you need to convert the values to character. What is the purpose? > > It's clear that if I convert element by element this > will take lot of time in my macro ! It should not be that slow. Assuming you need to do it at all. > > Data-null , I've tested your program but it shows > many errors ! Worked very well for me. It is a PROC IML program, so I suspect you don't have IML.

data_null__ · ‎08-24-2009

The syntax you show implies that you are using IML. [pre] proc iml; reset print; a={1,2,3}; b = char(a,1); quit; [/pre]

data_null__ · ‎08-23-2009

Is this what you mean? [pre] data delete; set sashelp.vmacro; where scope eq: 'G' and name ne: 'SYS'; run; data _null_; set delete; call symdel(name); run; [/pre]

data_null__ · ‎08-21-2009

If you use ABS(B-50) as a distance then find MIN would that be adequate? [pre] data new; input A B; dif = abs(b-50); datalines; 1 50.10 1 50.2 2 49.99 2 49.7 3 49.8 3 49.90 4 50.7 4 50.3 5 50.5 5 50.6 ; run; proc summary nway; class a; output out=work.closest(drop=_:) idgroup(min(dif) out(b)=); run; proc print; run; [/pre]

data_null__ · ‎08-21-2009

I don't know of an elegant way to evaluate the expression in KEY as the data step executes. SAS does not have an EXECUTE (data step code) function, AFAIK. You could code gen some "wall paper" for example. I hope my example data is suitable to model your process. [pre] data test; input (p1-p10)(1.) +1 key $64.; id + 1; cards; 0010011101 p7 and ( p3 or p8 ) 0010010101 p7 and ( p3 or p8 ) 0010010101 p4 and ( p5 or p8 ) 0010010101 p7 and ( p4 or p10) 0010010101 p6 and ( p3 or p8 ) ;;;; run; proc sort data=test(keep=key) out=keys nodupkey; by key; run; filename FT79F001 temp; data _null_; file FT79F001; set keys; put +6 'when(' key:$quote66. ') iscorrect = ' key ';'; run; data iscorrect; set test; select(key); %inc FT79F001 / source2; otherwise; end; run; proc print; run; [/pre]

data_null__ · ‎08-21-2009

> Unfortunately, I need to use this list to do things > like create define statements for proc report. I > really need the list stored in a macro variable that > can be referenced Unfortunate? What information do you need to supply for the DEFINE statements, you may be able to write generic define statements. Consider the following... [pre] proc sql noprint nowarn outobs=5 /*or 5*/; select name into :names separated by ' ' from dictionary.columns where libname eq 'SASHELP' and memname eq 'CLASS' order by varnum ; quit; run; %put NOTE: NAMES=&names; proc report data=sashelp.class(keep=&names) nowd list; columns &names; define _all_ / group; define _numeric_ / format=comma12.; define _character_ / format=$10.; run; [/pre] Or if that is inadequate you could "code gen" define statements in a way similar to the variable list &NAMES. A bit more complicated by doable.

data_null__ · ‎08-21-2009

You probably don't need to create a new data set just get the list of names. You want the 5 left most by position in the data set right? [pre] proc sql noprint nowarn outobs=3 /*or 5*/; select name into :names separated by ' ' from dictionary.columns where libname eq 'SASHELP' and memname eq 'CLASS' order by varnum ; quit; run; %put NOTE: NAMES=&names; proc report data=sashelp.class nowd; columns &names; run; [/pre]

data_null__ · ‎08-21-2009

'week.3' [pre] data _null_; do d = today() to today()+14; t = intnx('week.3',d,1); put (d t) (+5 weekdate.-r); end; run; [/pre]

data_null__ · ‎08-21-2009

Hi Chris, I glad you liked the data step enough want to use it. Sorry I did not see the question regarding PUT _ALL_; vs PUT (_ALL_)(:); sooner. I think the question has been answered but I will give you my "explanation". Hopefully it wont be more confusing or wrong. PUT _ALL_; is a special put statement specification. DUMP the PDV, including _ERROR_ and _N_ and the first and last dot variables and any others I forgot to mention. As far as I know the destination is not important, i.e. FILE LOG or FILE other. The location is not important as PUT _ALL_; in this context means everything in the PDV. _ALL_ is also a specification for a SAS Variable List. When used in PUT the parenthesis distinguish the two different functions, very subtle you must admit. I believe the thread has a reference to this powerful SAS feature. SAS Variable Lists are a key feature of dynamic programming. When used in a data step put (_ALL_)( ) list is ,all the variables defined when the _ALL_ is encountered as the data step is being compiled. Placement of RETAIN or any other statements that define variables is important to determine which variables are included in the (_ALL_) list. PUT (_ALL_)(:); can produce different results when used in different locations within a data step. PUT _ALL_; always puts variables in the PDV. [pre] 15 data _null_; 16 put 'NOTE: LIST ' (_all_)(:); 17 put 'NOTE- PDV ' _all_; 18 set sashelp.class; 19 put 'NOTE: LIST ' (_all_)(:); 20 put 'NOTE- PDV ' _all_; 21 stop; 22 run; NOTE: LIST PDV Name= Sex= Age=. Height=. Weight=. _ERROR_=0 _N_=1 NOTE: LIST Alfred M 14 69 112.5 PDV Name=Alfred Sex=M Age=14 Height=69 Weight=112.5 _ERROR_=0 _N_=1 NOTE: There were 1 observations read from the data set SASHELP.CLASS. [/pre] The way the (_ALL_) list is defined is why my data step uses the LINK statement so the helper variables do not become part of put (_ALL_)( ); You can use PUT (_ALL_)(:) at various locations within a data step a way to "study" what variables are defined at a particular location. Could be useful perhaps. Let me know if you have other questions.

data_null__ · ‎08-20-2009

> Sort by subject treatment dateprescribed. > > data xxx; > by subject treatment dateprescribed; > if first.treatmant; > run; This wont work because the data need to be ordered by date, not treatment. But, the FIRSTing needs to be done by TREATMENT. The NOTSORTED by statement option is the key. [pre] data rx; input (subject treatment dateprescribed)(2*:$1. :mmddyy.); format date: mmddyy10.; cards; 1 A 06/10/2002 I mixed up the lines a bit 1 A 09/11/2003 1 A 08/08/1997 1 B 07/09/2000 1 B 05/09/2001 1 A 05/06/2002 1 A 09/12/1998 1 A 12/01/1999 2 A 08/03/1996 2 A 09/12/1998 2 B 12/01/1999 2 B 07/09/2000 2 B 05/09/2001 2 A 05/06/2002 2 A 06/10/2002 2 B 09/11/2003 ;;;; run; proc sort; by s: d:; run; proc print; by s: t: notsorted; id s: t:; var d:; run; data firstInSequence; set; by s: t: notsorted; if first.treatment; run; proc print; by s:; id s:; run; [/pre]

data_null__ · ‎08-20-2009

You "can" get proc freq to display rows and columns of zeros but it has to get some help from another source. Using PROC SUMMARY with PRELOADFMT is one way to help proc freq. Note ZEROS options on weight is version V9 and up. [pre] proc format; value all 1=1 2=2 3=3; run; data zeros; output; call missing(row,col); run; proc summary nway completetypes; class row col / preloadfmt; format row col all.; output out=counts(drop=_type_); run; proc freq; tables row*col; weight _freq_ / zeros; format row col; run; [/pre]

data_null__ · ‎08-20-2009

filename dummy dummy; proc printto log=dummy; run; [pre] device-type specifies the type of device or the access method that is used if the fileref points to an input or output device or location that is not a physical file: DUMMY specifies that the output to the file is discarded. Tip: Specifying DUMMY can be useful for testing. [/pre]

Online Status	Offline
Date Last Visited	11 hours ago