On 03/07/13 04:21, Venkatesh Nagarajan wrote:
> I am trying to understand lagged correlations.
>
> x= 1:100;
> y = c(rep(NA,40), 1:60)ccf(x = x, y = y, lag.max=100, na.action=na.pass, type
> = "correlation")
>
> I was hoping to see max cor at lag = 40. But I am not. What am I doing wrong?
Well, I would have expected the correlation to be equal to 1, at any
(meaningful) lag.
Essentially the idea is cor(x,x+a) = 1 for any constant a.
Experimenting with cor(...,use="pair") and various lags would seem to
bear this out.
Note that the help for acf/ccf says:
> The lag |k| value returned by |ccf(x, y)| estimates the correlation
> between |x[t+k]| and |y[t]|.
E.g. lag 40 (cor(x[t+40],y[t]):
cor(x[41:140],y,use="pair") # Yields 1.
E.g. lag -40 (cor(x[t-40],y[t]):
cor(c(rep(NA,40),x[1:60]),y,use="pair") # Yields 1.
So I am mystified by the output of ccf().
Perhaps someone would care to explain .....
Or perhaps not. :-)
cheers,
Rolf Turner
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
