Date:         Fri, 5 Sep 2003 10:21:04 +0200
Reply-To:     "Groeneveld, Jim" <jim.groeneveld@VITATRON.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "Groeneveld, Jim" <jim.groeneveld@VITATRON.COM>
Subject:      Re: Help with a loop!
Comments: To: Roberto Valdivia <valdivia@MONTANA.EDU>
Content-Type: text/plain; charset="iso-8859-1"

Hi Roberto,

No other replies to your Q yet, so it is my turn [;-)

For TN (or _N_) = 1 you already indicate DPC and DCQ to be missing. So in that case CTi would be missing as well. That means that CTAi would be missing for TN=1 too and while all subsequent CTAi's would be accumulated from CTAi for TN=1, they all would be missing. I assume that you want all non-missing sums, so I have developed the following code:

DATA LoopTest;
  INPUT TN PC CQ;
  DPC=PC-LAG1(PC);
  DCQ=CQ-LAG1(CQ);
  CTi=dpc*dcq/2;
  RETAIN CTAi;
  IF (_N_ /* or TN */ EQ 1) THEN CTAi = CTi; * thus missing for TN=1;
  ELSE CTAi = SUM ( CTi, CTAi );
* Even without the IF/ELSE the last assignment could apply always, also for TN=1;
  CARDS;
1   25   245
2   50   543
3   75   645
4  100  725
5  125  876
6  150  965
7  175 1145
8  200 1178
9  225 1178
10 250 1178
;
RUN;

PROC PRINT DATA = LoopTest; RUN;

Regards - Jim.

Y. (Jim) Groeneveld MSc
Biostatistician
Vitatron B.V.
Meander 1051
6825 MJ Arnhem
The Netherlands
+31/0 26 376 7365; fax 7305
Jim.Groeneveld@Vitatron.com
www.vitatron.com


-----Original Message-----
From: Roberto Valdivia [mailto:valdivia@MONTANA.EDU]
Sent: Thursday, September 04, 2003 23:13
To: SAS-L@LISTSERV.UGA.EDU
Subject: Help with a loop!
Importance: High


Hi!, I have a problem writting a code to make an analysis, it might be
simple to solve,
but unfortunately, I couldn't get it right. So, please, could anyone help me
with this?.
Below is the problem described,

Thanks in advance,

Roberto.


I have a data like this:

tn  pc  cq
1   25   245
2   50   543
3   75   645
4  100  725
5  125  876
6  150  965
7  175 1145
8  200 1178
9  225 1178
10  250 1178

First I need to estimate the difference between PCi - PCi-1 (where i=tn,
i.e. when TN=2 and when TN=1 => 50-25=25),
I did this using the "lag" command :
DPC=PC-LAG1(PC);  Same thing for the CQ variable:
DCQ=CQ-LAG1(CQ);

Now I have these two variables in the dataset (note that the first
observation is missing for DPC and DCQ):

tn  pc  cq   dpc dcq
1  25   245    .   .
2  50   543   25 298
3  75   645   25 102
4 100  725   25  80
5 125  876   25 151
6 150  965   25  89
7 175 1145  25  180
8 200 1178   25  33
9 225 1178   25   0
10 250 1178 25   0

Here is my question:

I need to estimate a value like CTi=dpc*dcq/2 (i is the observation number).
After this, I need to have a variable that accumulate this value
each "tn", so for example: If "i" is the index for TN (observation number)
for TN=1 CTAi=CTi
for TN=2 CTAi=CTi + CTi-1
for TN=3 CTAi=CTi + CTi-1 +CTi-2

 OR, IN GENERAL:  CTAi=CTi+CTAi-1

I've tried to do this using the DO statement to build a loop but I cannot
get it work.

Any help will be appreciated.