Skip to content

Getting error message using R: Please help (it's coursework due in on Friday)

9 messages · ethanhaas, Peter Dalgaard, Gavin Simpson

#
I've been trying for the past 3 weeks to use R (much better than Matlab but I
am very bad with computers so I am very new to all of this) and know how to
input the data (hey, it's a start!) but every time I type in the following:

dcm <- decompose(information)

I get the following error message:

Error in decompose(information) : time series has no or less than 3 periods

Could you please aid me so that I can progress to the next step. What I am
in fact trying to do at the moment is plot a graph of the information so I
had assumed it would be:

dcm <- decompose(information)
Plot(dcm$trend)

but obviously it does not appear to be the case.

By the way the code I have right now in case it helps is:

information <- scan("everything.txt")
timeseries <- ts(information, frequency = 11, start = c(1970, 1994))
using <- window(timeseries, end = c(1990, 11))
dcm <- decompose(information)
Plot(dcm$trend)


Thank you very much in advance, I really appreciate the help.
#
ethanhaas wrote:
Er, are you sure it is "information" that you want to decompose()? (If
so, what are the computations of "timeseries" and "using" supposed to be
good for?)

  
    
#
I simply called it information but I can also call it data if you wish, it
doesn't make a difference.
The timeseries is simply the time series of this data between the two dates
and using is the data up to 1990 as opposed to 1994.
P.Dalgaard wrote:

  
    
#
ethanhaas wrote:
And isn't it one of the latter two you want to decompose()?

  
    
#
What I'm trying to do is have plots and summary statistics of x
(autocorrelations for example) where x refers to the data in using, and I
thought decomposing would be the way to go about it.
P.Dalgaard wrote:

  
    
#
ethanhaas wrote:
You are not listening. I give up.

  
    
#
I was, and the answer is yes, but I figured the easiest way for you to
understand is for me to say exactly what I am trying to do. Perhaps I don't
even need to decompose (although I think I do). If I can get past this
problem it would make everything else so much easier.
P.Dalgaard wrote:

  
    
#
On Wed, 2007-12-12 at 05:03 -0800, ethanhaas wrote:
Peter was trying to get you to see that you should be doing one of the
following:

dcm <- decompose(timeseries)

or

dcm <- decompose(using)

*Not* decompose(information), which will try to decompose your data
frame, which is not a timeseries object in the sense of ?ts and
therefore doesn't have periods and the like, hence the error.

Do either of the two suggestions that Peter clearly suggested you try
(and I repeat above) work for you and give you what you want?

HTH

G
#
Although I was wrong on what I was trying to decompose that didn't work
either however after a conversation with the lecturer the problem has been
found and solved.
Gavin Simpson wrote: