[R] how to extract data by specific months from a monthly ts object

[Marc Schwartz]

tom soyer wrote:
> Hi,
> 
> I am trying to extract data from a ts object by month, e.g., extract Jan,
> Feb, and Aug data from a monthly ts object. I tried the following but it
> didn't work:
> 
>> xa=1:50
>> ta=ts(xa,start=c(1990,1),frequency=12)
>> ta[cycle(ta)==c(1,2)] # this method works but it's not what I want
>  [1]  1  2 13 14 25 26 37 38 49 50
>> ta[cycle(ta)==c(1,2,8)] # this method doesn't work, unfortunately
>  [1]  1  2 13 14 25 26 37 38 49 50
> Warning message:
> In `==.default`(cycle(ta), c(1, 2, 8)) :
>   longer object length is not a multiple of shorter object length
> 
> somehow ta[cycle(ta)==c(1,2)] works. But what I really need is for
> ta[cycle(ta)==c(1,2,8)] to work, and it doesn't. Does anyone know how could
> to do this kinds of data extraction?
> 
> Thanks!

Tom,

You need to use %in% when you want to test whether one or more values
are in a second set of values.

In this case:

> ta[cycle(ta) %in% c(1, 2, 8)]
 [1]  1  2  8 13 14 20 25 26 32 37 38 44 49 50



BTW, the use of:

> ta[cycle(ta) == c(1, 2)]
 [1]  1  2 13 14 25 26 37 38 49 50

results in a correct response, but largely by chance in this case
because of the way in which the data is stored. You should be using:

> ta[cycle(ta) %in% c(1, 2)]
 [1]  1  2 13 14 25 26 37 38 49 50


Note however, that if we use:

> ta[cycle(ta) == c(1, 3)]
[1]  1 13 25 37 49

we do not get the correct answer, which should be:

> ta[cycle(ta) %in% c(1, 3)]
[1]  1  3 13 15 25 27 37 39 49


See ?"%in%" for more information.

HTH,

Marc Schwartz