[R] subsetting, aggregating and zoo

[antonio rodriguez]

Gabor Grothendieck escribió:
> Try this:
>
> # test data
> x <- c(1:4, 6:8, 10:14)
> z <- zoo(x, as.Date(x))
>
> # idx is 1 for first run, 2 for second run, etc.
> idx <- cumsum(c(1, diff(z) != 1))
>
> # starts replaces each time with the start time of that run
> # ends is similar but for ends
> starts <- time(z)[match(idx, idx)]
> ends <- time(z)[cumsum(table(idx))[idx]]
>
> # average over each run using the time of the end of run for the result
> # replace ends with starts if that is preferred
> aggregate(z, ends, mean)
>
Yes it's OK in your example, but when I try to do it with my data I 
don't get the same figure.

is.zoo(z)
[1]TRUE

atributes(z)
$index
   [1] "1988-01-13" "1988-01-14" "1988-01-16" "1988-01-20" "1988-01-21"
..................................................................................................
[3861] "2005-12-20" "2005-12-23" "2005-12-24" "2005-12-25" "2005-12-26"
[3866] "2005-12-27" "2005-12-30"

$class
[1] "zoo"

z[1:10]

1988-01-13 1988-01-14 1988-01-16 1988-01-20 1988-01-21 1988-01-22 1988-01-25
  7.973946   9.933518   7.978227   7.512960   6.641862   5.667780   5.721358
1988-01-26 1988-01-27 1988-01-28
  6.863729   9.600000   9.049846

If I follow your instructions,

idx <- cumsum(c(1, diff(z) != 1))
starts <- time(z)[match(idx, idx)]
ends <- time(z)[cumsum(table(idx))[idx]]

s1 <- aggregate(z, starts, mean)
s1[1:10]

1988-01-13 1988-01-14 1988-01-16 1988-01-20 1988-01-21 1988-01-22 1988-01-25
  7.973946   9.933518   7.978227   7.512960   6.641862   5.667780   5.721358
1988-01-26 1988-01-27 1988-01-28
  6.863729   9.600000   9.049846

s2 <- aggregate(z, starts, mean)
s2[1:10]

1988-01-13 1988-01-14 1988-01-16 1988-01-20 1988-01-21 1988-01-22 1988-01-25
  7.973946   9.933518   7.978227   7.512960   6.641862   5.667780   5.721358
1988-01-26 1988-01-27 1988-01-28
  6.863729   9.600000   9.049846


Always the same. Don't know why (there are not NA's in the series)

Antonio