--- David L Cassell <davidlcassell@MSN.COM> wrote:

> topkatz@MSN.COM wrote: > >I'm trying to do a basic task in IML and failing miserably. This is > a > >true "RTFM" question, so if that offends you, please read no > further. > > > >I want to read a data set into IML, run a procedure, and put the > output > >into a data set. Here is what happened: > >

----------------------------SNIP----------------------------

> > > >NOTE: Initial allocation of symbol space exhausted. You may specify > >SYMSIZE= option on PROC IML > > statement to increase its allocation for more efficiency. > >1671 > >1672 quit ; > >NOTE: Exiting IML. > >NOTE: The data set DISTOUT has 0 observations and 10000 variables. > >

----------------------------SNIP----------------------------

> > > >How can I get what I want?

> > It looks to me like you're not failing miserably, just not > quite getting results. There is a difference. > > As I see it, you have two options. I prefer #2, but it is up > to you. > > [1] Increase SYMSIZE as much as your OS will let you. > Normally, your symbol space will grow as you need it. And > normally, when you only have a couple matrices to muck > with, you're more likely to need to enlarge your WORKSIZE= > number. You can start by using the SHOW SPACE command in > IML to see what the settings on SYMSIZE and WORKSIZE are. > A lot of the modules are dynamically loaded, so don't > take up all the memory or you may not be able to load the > MCD module! You may have to enlarge both SYMSIZE and > WORKSIZE to get your process to run. IML is memory-intensive. > > [2] Or you can scrap the IML code and do this using PROC > ROBUSTREG. >

Talbot,

If you wish to to stay within IML for whatever reason, then you might want to construct a small data set of, say, 1000 observations and 30 or 40 variables. Try running your code against that data. Do you get results which you expect? Or does the program still fail?

I have not used the MCD call routine, and don't have time to RTFM. But perhaps the suggestion to pare down the data will help you identify whether the problem is with your code or whether the problem is with the size of your data.

Dale

--------------------------------------- Dale McLerran Fred Hutchinson Cancer Research Center mailto: dmclerra@NO_SPAMfhcrc.org Ph: (206) 667-2926 Fax: (206) 667-5977 ---------------------------------------

__________________________________________ Yahoo! DSL – Something to write home about. Just $16.99/mo. or less. dsl.yahoo.com