Skip to content

simulation study using R

6 messages · Davood Tofighi, jim holtman, Paul Gilbert

#
What is the format of the data you are storing (single value,
multivalued vector, matrix, dataframe, ...)?  This will help formulate
a solution.  What do you plan to do with the data?  Are you going to
do further analysis, write it to flat files, store it in a data base,
etc.?  How big are the data objects you are manipulating?
On Mon, Mar 3, 2008 at 7:05 PM, Davood Tofighi <dtofighi at asu.edu> wrote:

  
    
#
One of the things you might take a look at is the 'filehash' package.
It is an easy way of storing/retrieving R objects.  I have an
application where my objects are matrices of about the same size and I
can quickly store the data and then come back later with a different
script to do further analysis.
On 3/3/08, Davood Tofighi <dtofighi at asu.edu> wrote:

  
    
1 day later
#
Davood Tofighi wrote:
I generally find it is better to store the seed and other data you need 
to reproduce the experiment, rather than trying to store the data (see, 
for example, package setRNG). If you save only the summary statistics 
you need, then you can usually do it in memory. (Be sure to assign 
variables for the statistics to their full size first and then populate 
them, rather than extending them at each step.) If you write things to 
files then it will slow down your simulation a lot. In fact, in most 
cases it will be quicker to re-run the experiment than it will be to 
read the data from disk.

Paul
====================================================================================

La version fran?aise suit le texte anglais.

------------------------------------------------------------------------------------

This email may contain privileged and/or confidential in...{{dropped:26}}