I recommend the SAS product documentation if you encounter a new function, that part of it is perhaps the most useful :)

In this case, whichn selects one of N values based on the value (presumably in a variable) provided in the first parameter.

So:

x=4; y = whichn(x, 1,4,9,16,25,36); will give you a value of 16 in y, while x=2; y = whichn(x, 1,4,9,16,25,36); will give you a value of 4 in y.

http://support.sas.com/documentation/cdl/en/lefunctionsref/63354/HTML/default/p0zs0pv38mel2jn1in4lte2akx4d.htm

WHICHC is the character version of WHICHN, so it still takes a number for the first parameter but it returns a character value.

-Joe

On Wed, May 9, 2012 at 9:24 AM, ethan zhou <e75wez1@gmail.com> wrote:

> Mike, > Your solution is so simple. I like it. > > But I'm not familiar the SAS function "whichn" . Could you please explain > it to me? > > Also in my real data set, the variable names are like " cfed11 cfed12 > cfed13, pct2 pct3 pct4" other than "product1 product2 product3, pct1 pct2 > pct3" with consecutive numerical suffix. > I wonder if one can use your code without renaming those variables to > "product1 product2 product3, pct1 pct2 pct3". > > Thanks for your helps. > > Ethan > On Tue, May 8, 2012 at 2:14 PM, bbser 2009 <bbser2009@gmail.com> wrote: > > > Below are two alternative algorithms to do this. > > The first one is using bubble sorting. > > The second one is based on the ad hoc method of sorting three values, > just > > for fun. > > > > Regards, Max > > (Maaxx) > > > > data have1 have2; > > input ID product1 $ product2 $ product3 $ pct1 pct2 pct3; > > cards; > > > > 1 C D A 20 50 30 > > 2 B A E 55 40 5 > > 3 A C B 0 0 100 > > 4 F E B 40 50 10 > > 5 A D E 40 20 40 > > ; > > > > ***Bubble Sorting; > > data have1; > > modify have1; > > array product{3} $ 1; > > array pct{3}; > > dim=dim(pct); > > do i=1 to dim-1; > > do j=2 to dim; > > pcti=pct{i}; > > pctj=pct{j}; > > if pcti<pctj then > > do; > > pct{j}=pcti; > > pct{i}=pctj; > > temp_product=product{i}; > > product{i}=product{j}; > > product{j}=temp_product; > > end; > > end; > > end; > > run; > > proc print;run; > > > > ***Ad hoc sorting; > > data have2; > > modify have2; > > array product{3} $ 1; > > array pct{3}; > > array pr{3} $ 1 ; > > array pc{3}; > > do i=1 to 3; > > pr{i}=product{i}; > > pc{i}=pct{i}; > > end; > > pct1=max(of pc{*}); > > pct3=min(of pc{*}); > > pct2=100-pct1-pct3; > > do i=1 to 3; > > do j= 1 to 3; > > if pct{i}=pc{j} then > > do; > > product{i}=pr{j}; > > pc{j}=-1; > > leave; > > end; > > end; > > end; > > run; > > proc print;run; > > > > > > -----Original Message----- > > From: SAS(r) Discussion [mailto:SAS-L@LISTSERV.UGA.EDU] On Behalf Of > Ethan > > Zhou > > Sent: May-07-12 1:46 PM > > To: SAS-L@LISTSERV.UGA.EDU > > Subject: [SAS-L] need helps for data mamipulation > > > > Hi all, > > > > I have a data with up to 3 different products associated with their > > percentages usage respectively. Those percentages add up to 100%. > > > > > > How to horizontally sort the product1, product2 and product3 based on > > their percentage value in descending order? > > > > Here is the sample data: > > > > data have; > > input ID product1$ product2 $ product3 $ pct1 pct2 pct3; > > cards; > > > > 1 C D A 20 50 30 > > 2 B A E 55 45 5 > > 3 A C B 0 0 100 > > 4 F E B 40 50 0 > > 5 A D E 40 20 40 > > ; > > > > The wanted results: > > Id product1 product2 product3 pct1 pct2 pct3 > > 1 D A C 50 30 20 > > 2 B A E 55 45 5 > > 3 B A C 100 0 0 > > 4 E F B 50 40 0 > > 5 A E D 40 40 20 > > > > > > > > Thank you so much for your help. > > > > John > > >