I think reading 20,000+ files with PROC IMPORT will be too slow. 1) Can the files be read using the same input statements. Hopefully you don't have 20,000 file with vastly different record layout. 2) show directory and file name "structure" in more detail. You description was too abstract for me, I'm not very bright. 3) I didn't get that bit about 3 of 7 or something and how that is determined. 4) This call probably all be done with ONE data step. But I need to know more details. The "have" and "need" so to speak. ... View more

I would try TRUNCOVER INFILE statement option. ... View more

To put it another way you want to KEEP the last record.; [pre] data table; do var1=1 to 10; var2 = ranuni(1); output; end; run; data last; do point=nobs; set table point=point nobs=nobs; output; end; stop; run; proc print; run; [/pre] ... View more

[pre] ods csv file='csvWithLabels.csv'; proc print label noobs data=sashelp.shoes(obs=20); run; ods csv close; [/pre] ... View more

I think the point is to calculate the sum. [pre] data have; input Id:$1. Amount; cards; Q 1200 W 1500 E 1100 ;;;; run; data sum; length ID $3; * "Sum" won't fit in $1; do while(not eof); set have end=eof; output; sum + amount; end; id = 'Sum'; amount = sum; output; stop; drop sum; run; proc print; run; [/pre] ... View more

> Thanks d_n, this is actually a hobby gone slightly > out of control. > The problem is whether CO2 varies with time and > position in an aquarium regulated with a controller. > I reported the data as indicated in the table. Some > one else used the data to do the analysis and create > the outputs. > he problem is that the data is not normal because it > has been skewed using a non linear transformer > (exponentiation). When asked to redo the analysis > using the untransformed data, the person refused so > now I have to do it myself. > > As I said, I don’t know anything about SAS but the > code reminds me of Basic. Really. > > I cannot run the program now since I’m not at school > but I have some questions. > > Did you run the program? Does it produce the same > two outputs? yes exactly the same as you posted. [pre] The SAS System 58 13:15 Friday, February 27, 2009 Obs position day co2 1 1 1 19 2 1 2 19 3 1 3 20 4 1 4 21 5 1 5 17 6 2 1 23 7 2 2 23 8 2 3 20 9 2 4 23 10 2 5 19 11 3 1 24 12 3 2 21 13 3 3 21 14 3 4 23 15 3 5 17 16 4 1 22 17 4 2 21 18 4 3 20 19 4 4 21 20 4 5 18 21 5 1 21 22 5 2 19 23 5 3 21 24 5 4 23 25 5 5 17 26 6 1 24 27 6 2 23 28 6 3 20 29 6 4 23 30 6 5 21  The SAS System 59 13:15 Friday, February 27, 2009 The GLM Procedure Class Level Information Class Levels Values position 6 1 2 3 4 5 6 day 5 1 2 3 4 5 Number of Observations Read 30 Number of Observations Used 30  The SAS System 60 13:15 Friday, February 27, 2009 The GLM Procedure Dependent Variable: co2 Sum of Source DF Squares Mean Square Model 9 97.6666667 10.8518519 Error 20 25.1333333 1.2566667 Corrected Total 29 122.8000000 Source F Value Pr > F Model 8.64 <.0001 Error Corrected Total R-Square Coeff Var Root MSE co2 Mean 0.795331 5.389478 1.121011 20.80000 Source DF Type I SS Mean Square position 5 29.20000000 5.84000000 day 4 68.46666667 17.11666667 Source F Value Pr > F position 4.65 0.0056 day 13.62 <.0001 Source DF Type III SS Mean Square position 5 29.20000000 5.84000000 day 4 68.46666667 17.11666667 Source F Value Pr > F position 4.65 0.0056 day 13.62 <.0001  The SAS System 61 13:15 Friday, February 27, 2009 The GLM Procedure t Tests (LSD) for co2 NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate. Alpha 0.05 Error Degrees of Freedom 20 Error Mean Square 1.256667 Critical Value of t 2.08596 Least Significant Difference 1.4789 Means with the same letter are not significantly different. t Grouping Mean N position A 22.2000 5 6 A B A 21.6000 5 2 B A B A 21.2000 5 3 B B C 20.4000 5 4 B C B C 20.2000 5 5 C C 19.2000 5 1  The SAS System 62 13:15 Friday, February 27, 2009 The GLM Procedure t Tests (LSD) for co2 NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate. Alpha 0.05 Error Degrees of Freedom 20 Error Mean Square 1.256667 Critical Value of t 2.08596 Least Significant Difference 1.3501 Means with the same letter are not significantly different. t Grouping Mean N day A 22.3333 6 4 A A 22.1667 6 1 A B A 21.0000 6 2 B B 20.3333 6 3 C 18.1667 6 5 [/pre] > > If I change the size of the array, how do I modify > the code to account for this? Array? Number of Positions? Number of Days? Not sure what you are referring too. To me array is not the "correct" term here, but that can be resolved easily enough. If you can post or describe the new data I'm sure "we", the discussion group, can help you. > > Thanks again for your help. I’ll head back to school > tomorrow and try to run the program. > RTP ... View more

I don't understand your question. If you have the output way do you need the code to produce it? Also this isn't cluster analysis. [pre] data co2; input position @; do day = 1 to 5; input co2 @; output; end; cards; 1 19 19 20 21 17 2 23 23 20 23 19 3 24 21 21 23 17 4 22 21 20 21 18 5 21 19 21 23 17 6 24 23 20 23 21 ;;;; run; proc glm data=co2; class position day; model co2 = position day; means position day / lsd lines; run; quit; [/pre] ... View more

> They should be presented in alphabetical order. Sorry I missed that part. [pre] proc plan; factors n=1000 ordered color = 1 of 12 / noprint; output out=colors color cvals=("BLACK" "WHITE" "RED" "GREEN" "BLUE" "CYAN" "MAGENTA" "GRAY" "PINK" "ORANGE" "BROWN" "YELLOW"); run; proc summary data=colors nway order=freq; class color; output out=freq(drop=_type_); run; proc print; run; proc format; value $color(default=12) 'ff'x='Other'; run; data freqV / view=freqV; set freq; if _n_ gt 3 then color='ff'x; run; proc summary data=freqV nway order=internal; class color; format color $color.; output out=freq4(drop=_type_); freq _freq_; run; proc print; run; Obs color _FREQ_ 1 BLACK 100 2 BROWN 100 3 WHITE 94 4 Other 706 [/pre] ... View more

There may be better way but you need to count the colors then pool the ones that are not in the top three. Notice the ORDER= options in the PROC SUMMAY steps that's what makes it work. It gives a different answer each time it is run. [pre] proc plan; factors n=1000 ordered color = 1 of 12 / noprint; output out=colors color cvals=("BLACK" "WHITE" "RED" "GREEN" "BLUE" "CYAN" "MAGENTA" "GRAY" "PINK" "ORANGE" "BROWN" "YELLOW"); run; proc summary data=colors nway order=freq; class color; output out=freq(drop=_type_); run; proc print; run; data freqV / view=freqV; set freq; if _n_ gt 3 then color='Other'; run; proc summary data=freqV nway order=data; class color; output out=freq4(drop=_type_); freq _freq_; run; proc print; run; [/pre] ... View more

ANYDIGIT function would probably be faster. If you are just looking at the first position. The technique you are using is better for test if a string is a number, so strings like 1e4 would be read correctly. If your numbers could include commas or dollar sign use COMMA informat. If you are trying to see if string is a SAS name NVALID function. [pre] data _null_; input string $16.; x = anydigit(string); if anydigit(string) eq 1 then put 'Digit in First Position' string=; put (_all_)(=); cards; 1e000 e1000 .This This ;;;; run; [/pre] ... View more

Use ?? in format field of INPUT function. You may want to look a VERIFY function also. > I'm using this statement to get back the first > character in an address field. > I'm having a problem finding a way to verify what it > is. > char1=substr(addr,1,1); > > I tried this : > > if (input(addr, ??1.) eq .) then > a='non-numeric'; > e > a ='numeric'; > hen I try this I get a lot of > "NOTE: Invalid argument to function INPUT at line > 1063 column 9." How can I get it to have a clean > execution? > > Thank You ... View more

> Nice macro thou'.....isn't it...:) Well you don't need a macro to do this and if you did I would not create all those macro variables. What your example does remind me of is the need for PROC CONTENTS as it provides alphabetical order by default, (you didn't need to sort), with special "attention" to SAS enumerated variable lists. (a1 followed by a2 not a10). This ordering is not obtained by a simple sort of the NAMES as in my first example using DICTIONARY.COLUMNS. See new example. [pre] data test; length b a g d 8 a4-a11 a1-a3 8; b = 1; a = 2; d=10; g = 30;output; call missing(of _all_); b = 2; g = 31; output; call missing(of _all_); b = 3; a = 9; d=7; g = 32;output; call missing(of _all_); b = 4; g = 33; output; run; proc print; run; proc contents data=work.test out=work.names(keep=name); run; proc sql noprint; select name into :vars seperated by ' ' from work.names ; quit; run; %put NOTE: vars=&vars; data test2; retain &vars; set test; run; proc contents varnum; run; Variables in Creation Order # Variable Type Len 1 a Num 8 2 a1 Num 8 3 a2 Num 8 4 a3 Num 8 5 a4 Num 8 6 a5 Num 8 7 a6 Num 8 8 a7 Num 8 9 a8 Num 8 10 a9 Num 8 11 a10 Num 8 12 a11 Num 8 13 b Num 8 14 d Num 8 15 g Num 8 [/pre] ... View more

> Your example works if all obs has values on every > var, try this: > data test; > length b a g d 8; > b = 1; a = 2; d=10; g = 30;output; > b = 2; g = 31;output; > b = 3; a = 9; d=7; g = 32;output; > b = 4; g = 33;output; > un; > > data test2; > retain a b d g; > set test; > run; > proc print data = test2;run; > > retain is used for retaining vars....I think...:-) This is a common misconception. Your example is flawed but I have corrected it below to show that using RETAIN is "safe" for reordering variables read with a SET statement. RETAIN is for retaining values but RETAIN does not have any effect on the values of variables read using a SET statement other than to reorder them when RETAIN is placed before the SET statement. [pre] data test; length b a g d 8; b = 1; a = 2; d=10; g = 30;output; call missing(of _all_); b = 2; g = 31; output; call missing(of _all_); b = 3; a = 9; d=7; g = 32;output; call missing(of _all_); b = 4; g = 33; output; run; proc print; run; data test2; retain a b d g; set test; run; proc print; run; **Output from second PROC PRINT; Obs a b d g 1 2 1 10 30 2 . 2 . 31 3 9 3 7 32 4 . 4 . 33 [/pre] ... View more

Using PROC FREQ the syntax will be more concise. [pre] ods select Freq.NLevels; ods output Freq.NLevels=Nlevels; proc freq data=sample levels; run; ods select all; Proc print; run; [/pre] ... View more

You don't want to use a LENGTH statement because you don't need to know those attributes and you don't want to fiddle them. Too much trouble. You just need alphabetical list and RETAIN. [pre] proc sql noprint; select name into :vars separated by ' ' from dictionary.columns where libname eq 'SASHELP' and memname eq 'CLASS' order by name ; quit; run; %put NOTE: VARS=&vars; data work.class; retain &vars; set sashelp.class; run; proc print; run; [/pre] ... View more