Re: Maximum number of cases SPSS-PC (fwd)

Thread View: bit.listserv.spssx-l

2 messages

[T] Back to Threads [V] Tree View [G] Group Articles

2 total messages Started by Mike Palij Thu, 22 Jul 1993 12:36

#294

Author: Mike Palij
Date: Thu, 22 Jul 1993 12:36

44 lines
2591 bytes

Bruce Peterson <brucep@SPSS.COM> writes:
>The number of cases that SPSS/PC+ and SPSS for Windows (and our
>other products) is limited only by the amount of available
>disk space.

I believe that Bruce and others who have provided this
bit of information may think that they're being helpful
but they're not.  The info given above does not tell a
user what the largest job (cases by vars) is that they can
run given the resources currently on their machine.  More
useful info could take one of the following forms:

(a) a formula that allows one to calculcate the resources
neccesary for a specific problem/procedure.  BMDP has
traditionally provided such formulas for some of their
programs.

(b) a table which allow one to look-up the minimum amount
of memory (real or virtual) neccesary for a job.  The
table would look something like this:

                     # of Vars
# of Cases    25        50        100        500         1000
    50      base.1    base.2     base.2     base.3      base.4
   100
   500
  1000
  2000
  5000
 10000
 15000

The cells would contain a base or minimum memory requirement
which could then be multiplied by a procedure "factor" to
identify how much memory a specific procedure requires (e.g.,
how much memory does regression require = base.# * regression
factor (1.5)).  This would allow one to determine whether one's
current dataset could be analyzed as is or if its necessary
to eliminate vars or cases.  This would be particularly useful
for procedures like Examine or Manova which choke on large
problems (at least the SPSS/PC chokes).  BMDP also provides
a similar set-up for some of its programs.

**Mike Palij/Psychology Dept/New York University

Re: Maximum number of cases SPSS-PC (fwd)

#295

Author: Rauno Veijola
Date: Thu, 22 Jul 1993 20:11

24 lines
1632 bytes

Hi,

My two cents to the number of cases problem:

In SPSS, if you can get any version of the algorithms manual
(sorry I don't remember the name correctly), you can soon
find out that most of the procs are programmed so the
data needs to be gone thru only once and only one case
needs to be held in RAM at the same time. This is made possible
by using cumulative sums in the algorithms quite effectively.

So, for most of the procs, there is not any kinds of limits
for number of cases.  Or, in other words, the amount of memory
used for some proc doesn't depend on the number of cases.

Having PC's in question, the only limiting resource seems to
be the patience of the user. (Except for some special procs
like cluster analysis where one really needs to collect the whole
data to the RAM at the same time.)

Rauno Veijola
Computing Center,
University of Jyvaskyla, Finland
veijola@jyu.fi

Thread Navigation

This is a paginated view of messages in the thread with full content displayed inline.

Messages are displayed in chronological order, with the original post highlighted in green.

Use pagination controls to navigate through all messages in large threads.

Back to All Threads