R help - Mar 2006 - about pam

Hi there,

In the description of command "pam", it mentions "For datasets
larger
than (say) 200 observations". Now my dataset is a "54732 by 5"
dataframe
named "test". When I try to run pam(test,4),it shows " cannot
allocate
vector of length 1497768547". Is it because the row too big that it
can't handle?

Thank you!

>>>>> "Linda" == Linda Lei <llei at bccrc.ca>
>>>>>     on Thu, 16 Mar 2006 11:11:29 -0800 writes:
    Linda> Hi there,

    Linda> In the description of command "pam", it mentions
"For
    Linda> datasets larger than (say) 200 observations". Now my
    Linda> dataset is a "54732 by 5" dataframe named
    Linda> "test". When I try to run pam(test,4),it shows "
    Linda> cannot allocate vector of length 1497768547". Is it
    Linda> because the row too big that it can't handle?
yes.
You must need clara() instead of pam().

Though with today's fast computers, 
I'd advise to increase some of the defaults that made sense back
in the 80s: 'samples = 5' and 'sampsize' can easily be increased
by an order of magnitudes in order to get a more stable
(i.e. less random) result.
Also for your relatively large data matrix, you may want to use
'keep.data = FALSE'.

Regards,
Martin Maechler, ETH Zurich