Well,

I think Maarten is the best judge here.

On 6/16/06, toby dunn <tobydunn@hotmail.com> wrote: > Wensui , > > A few things strike me when I read your response. > > >>Well, the good side of my way is that you won't include the funky year you > >>don't want. > > I dont see this a good way to handle that. A where clause is more > appropriate solution to a 'Funky Year Problem'. Lets say for the moment I > was using your SQL solution, and lets say I have 50 years worth of data and > only one 'Funky Year Value'. I would have to copy and paste 100 lines of > code just to avoid that one 'Funky Year' Value. Now lets say I need to > rerun this again with 100 years worth of data, the code has to increase some > more. > > You see the problem isnt that funky year it should have been taken care of > before you started using the data set. The problem is the lack of > generality of the codes design to handle the problem. A good examle is the > set of programs I am finishing working on right now. The original set of > programs was around 20 to 25. Running some 40 or 50 thousand lines of > code. I rewrote the code in such a way as to concentrate on the business > rules instead of specific values. The end result was that the whole set of > programs went down to only 5 programs. Where it originally took 2 weeks to > make a run and check the results I now do it in 2 hours. By concentrating > on the business rules we found many errors that were present in the code > because I and the code reviewers were able to concentrat on the rules that > should govern how things were suppose to be done rather than the all the > values that where in the data set. Not only that but we have now expanded > the program suite to include full data profiles, create quality control > programs and data sets. Also, the general set up has allowed us to turn > around ad hoc requests in a few days rather than a few weeks. > > Now this new found speed and agility has made the bosses happy. They brag > about their abiity to turn around requests, and well other people not on my > project have started writing code in this manner which means that their turn > around time has gone down as well. In general it has had a pretty darn good > effect on the team as a whole. > > I think once you start coding in a more business rule oriented fashion you > will see the power and advantages that it has over a more data specific > approach. > > > Toby Dunn > > > > > > From: Wensui Liu <liuwensui@GMAIL.COM> > Reply-To: Wensui Liu <liuwensui@GMAIL.COM> > To: SAS-L@LISTSERV.UGA.EDU > Subject: Re: Percentage of subgroups > Date: Fri, 16 Jun 2006 09:09:37 -0400 > > Well, the good side of my way is that you won't include the funky year > you don't want. It is not surprise to have year "2049" in many > databases. > > On 6/16/06, data _null_; <datanull@gmail.com> wrote: > >When does an elegant solution involve mixing up the data with the code? > > > >What are you going to do when you get more years? > > > >On 6/15/06, Wensui Liu <liuwensui@gmail.com> wrote: > > > Maarten, > > > > > > Here is a more elegent solution in SQL: > > > > > > proc sql; > > > create table > > > wensui as > > > select > > > company, > > > sum(case when year = 2002 and delay ~= . then 1 else 0 end)/ > > > sum(case when year = 2002 then 1 else 0 end) as delay2002 format > >= percent8.2, > > > sum(case when year = 2003 and delay ~= . then 1 else 0 end)/ > > > sum(case when year = 2003 then 1 else 0 end) as delay2003 format > >= percent8.2 > > > from > > > test > > > group by > > > company; > > > quit; > > > > > > On 6/15/06, Havik, Maarten <Maarten.Havik@ordina.nl> wrote: > > > > Hi, > > > > > > > > Can someone please help me with a percentage question? > > > > I've tried a lot, and searched SAS-L but can't find the right posts.. > > > > I have a dataset like this: > > > > > > > > data test; > > > > input flight company $ year delay; > > > > datalines; > > > > 1 A 2002 13 > > > > 2 A 2002 . > > > > 3 A 2003 5 > > > > 4 A 2003 12 > > > > 5 A 2003 . > > > > 6 B 2002 17 > > > > 7 B 2002 28 > > > > 8 B 2003 32 > > > > 9 B 2003 . > > > > 10 B 2003 . > > > > 11 B 2003 . > > > > ; > > > > > > > > A missing for delay indicates a flight was on time. > > > > What i need is a report containing the percentage of delayed flights > >for each subgroup company & year, like this: > > > > > > > > Year > > > > Company 2002 2003 > > > > A 50% 66,66% > > > > B 100% 25% > > > > > > > > Any help is appreciated! > > > > Maarten > > > > > > > > > > > > > > > > > -- > > > WenSui Liu > > > (http://spaces.msn.com/statcompute/blog) > > > Senior Decision Support Analyst > > > Health Policy and Clinical Effectiveness > > > Cincinnati Children Hospital Medical Center > > > > > > > > -- > WenSui Liu > (http://spaces.msn.com/statcompute/blog) > Senior Decision Support Analyst > Health Policy and Clinical Effectiveness > Cincinnati Children Hospital Medical Center > > >

-- WenSui Liu (http://spaces.msn.com/statcompute/blog) Senior Decision Support Analyst Health Policy and Clinical Effectiveness Cincinnati Children Hospital Medical Center