Two things:
(1) type="l" won't work; plot.ecdf() calls plot.stepfun() and it
doesn't have a ``type'' argument.
(2) Giving plot two separate objects to plot makes no sense at all.
(3) If you've got ***frequency*** data then ecdf() is probably not
the right thing to do
in the first place.
(4) I can't count.
To get the ***plots** you want, do something like:
plot(ecdf(x.pre),do.points=FALSE,col="red",xlim=range(x.pre,x.obs))
plot(ecdf(x.obs),do.points=FALSE,col="blue",add=TRUE)
But as I said if x.obs and x.pre are counts, rather than the raw
data, this
does not appear to make any sense. You are getting the ecdf-s of the
counts
rather than of the raw data which is a toadally different story. And
toadally
misleading. Re-think.
cheers,
Rolf Turner
On 7/09/2009, at 1:52 PM, Roslina Zakaria wrote:
> Hi r-users,
>
> I would like to compare the cdf between historical and predicted.
> My x.obs and x.pre are the frequency data in classes of 0-300.
> I tried:
>
> plot(ecdf(x.obs),ecdf(x.pre),type="l",col="red")
>
> and it gives me:
>
> Error in plot.stepfun(x, ..., ylab = ylab, verticals = verticals,
> pch = pch) :
> argument 4 matches multiple formal arguments
>
> Thank you so much for any help given.
######################################################################
Attention:\ This e-mail message is privileged and confid...{{dropped:9}}