[FRIAM] R tutorial for idiots.

[Phil Henshaw]

Robert, 
Thanks for asking.  I'm getting a little help from Gabor on the use of
zoo, and will eventually learn how to code the things I want.   The R
core is really neat in that what would be 50 tangled lines of code in
lisp are simple short expressions in R.  That, however, means learning
lots and lots of objects and functions with complex properties and no
good reference.  It really is a sophisticated environment requiring high
level user knowledge.   Like I looked and looked tonight to find out how
to print one graph on top of another, or open multiple graph windows.
Couldn't find it.  
 
My other problem is the one that had me writing my own entire software
interface and analytic routines in the first place.  I seem to be using
curves for a different purpose than other people.    Say you have a
time-series of measures that reflect the life history of several
independent overlapping natural feedback loop systems, each with their
own separate growth, climax, disordering and decay periods. ¸¸¸¸.·´ ¯
`·.¸¸¸¸     You wouldn't want your way of rendering the record of those
events to erase all that critical beginning and ending information, and
paint the whole history with shapes displaying the same wiggly but
basically uneventful dynamics throughout.  That would make the
underlying systems much harder to find and figure out.
 
That's what I think splines do, for example, hide all the natural system
beginning and ending periods.      It may not be impossible to use
splines to faithfully represent underlying dynamics, but only the
rendering philosophy of their designers.    Of course stochastic and
simple differential equations are much worse.   On the other hand, I
don't know all the kinds of non-parametric methods of rendering
histories of events, but I get no hint from the discussion that anyone
is aware of this problem.   I can't find anyone who seems to be using
the shapes of their curves as a way of identifying and investigating the
behavior of natural feedback loop systems.   Everyone still seems to be
treating nature as if it should be following some formula, instead of
rendering events as part of local systems development.   
 
Do you know of anyone else with this complaint?

Phil Henshaw                       ¸¸¸¸.·´ ¯ `·.¸¸¸¸
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
680 Ft. Washington Ave 
NY NY 10040                       
tel: 212-795-4844                 
e-mail: pfh at synapse9.com          
explorations: www.synapse9.com <http://www.synapse9.com/>     

-----Original Message-----
From: friam-bounces at redfish.com [mailto:friam-bounces at redfish.com] On
Behalf Of Robert Holmes
Sent: Sunday, September 24, 2006 7:15 PM
To: The Friday Morning Applied Complexity Coffee Group
Subject: Re: [FRIAM] R tutorial for idiots.


Exactly what you want to do with these different timescales? If it's
just a case of sticking two lines on the same graph, that's simple. If
it's some multi-temporal statistic (is there such a thing?) that you
know of but isn't in R, then just code it up as an R function. 

Can you give us clarification? What are the statistic/statistics you
want to calculate to compare your time-series?

Robert  


On 9/23/06, Phil Henshaw <sy at synapse9.com> wrote: 

Who would guess just asking the R help forum would turn up a grad med
student whose PhD in modeling time series of epidemics with R, living
around the corner!     I guess I'm now officially one small step up from
rank beginner...    Anyway, it doesn't appear that R makes it at all
easy to use data with arbitrary time scales, however, which is
disappointing.    I just want a graphing database that other systems
researchers use that treats time as a real number.   Is that too much to
ask??     R's time-series objects appear to require orderly data with
points at constant time periods, and can't relate to others with
different periods...   My data and issues  tend to fail on both counts!

 
The session did give me a whole bunch of names of other people working
on non-parametric data imaging, so I'll bet someone somewhere deals
directly with the underlying problem, i.e. that life most certainly
happens, but just not on schedule!     I've gotten more than a couple
great suggestions on FRIAM!    Any others? 

 

Phil Henshaw                       ¸¸¸¸.·´ ¯ `·.¸¸¸¸



-------------- next part --------------
An HTML attachment was scrubbed...
URL: /pipermail/friam_redfish.com/attachments/20060924/793c599b/attachment.html