I have a problem where the endpoints() function in xts (and also the to.weekly function, which uses endpoints) sometimes returns Friday as the end of week, and sometimes returns Monday.
My dataset is called sp2.
> head(sp2)
[,1]
2012-01-09 1.78
2012-01-10 1.78
2012-01-11 1.77
2012-01-12 1.80
2012-01-13 1.77
2012-01-16 1.77
> tail(sp2)
[,1]
2012-06-28 1.94
2012-06-29 1.92
2012-07-02 1.92
2012-07-03 1.90
2012-07-04 1.89
2012-07-05 1.89
> class(sp2)
[1] "xts" "zoo"
> weekdays(index(sp2))
[1] "Monday" "Tuesday" "Wednesday" "Thursday" "Friday" "Monday" "Tuesday" "Wednesday" "Thursday" "Friday"
[11] "Monday" "Tuesday" "Wednesday" "Thursday" "Friday" "Monday" "Tuesday" "Wednesday" "Thursday" "Friday"
[21] "Monday" "Tuesday" "Wednesday" "Thursday" "Friday" "Monday" "Tuesday" "Wednesday" "Thursday" "Friday"
[31] "Monday" "Tuesday" "Wednesday" "Thursday" "Friday" "Monday" "Tuesday" "Wednesday" "Thursday" "Friday"
[41] "Monday" "Tuesday" "Wednesday" "Thursday" "Friday" "Monday" "Tuesday" "Wednesday" "Thursday" "Friday"
[51] "Monday" "Tuesday" "Wednesday" "Thursday" "Friday" "Monday" "Tuesday" "Wednesday" "Thursday" "Friday"
[61] "Monday" "Tuesday" "Wednesday" "Thursday" "Friday" "Monday" "Tuesday" "Wednesday" "Thursday" "Friday"
[71] "Monday" "Tuesday" "Wednesday" "Thursday" "Friday" "Monday" "Tuesday" "Wednesday" "Thursday" "Friday"
[81] "Monday" "Tuesday" "Wednesday" "Thursday" "Friday" "Monday" "Tuesday" "Wednesday" "Thursday" "Friday"
[91] "Monday" "Tuesday" "Wednesday" "Thursday" "Friday" "Monday" "Tuesday" "Wednesday" "Thursday" "Friday"
[101] "Monday" "Tuesday" "Wednesday" "Thursday" "Friday" "Monday" "Tuesday" "Wednesday" "Thursday" "Friday"
[111] "Monday" "Tuesday" "Wednesday" "Thursday" "Friday" "Monday" "Tuesday" "Wednesday" "Thursday" "Friday"
[121] "Monday" "Tuesday" "Wednesday" "Thursday" "Friday" "Monday" "Tuesday" "Wednesday" "Thursday"
So as you can see there are no missing weekdays in the data series.
Yet.....
> weekdays(index(to.weekly(sp2)))
[1] "Friday" "Friday" "Friday" "Friday" "Friday" "Friday" "Friday" "Friday" "Friday" "Friday" "Monday"
[12] "Monday" "Monday" "Monday" "Monday" "Monday" "Monday" "Monday" "Monday" "Monday" "Monday" "Monday"
[23] "Monday" "Monday" "Monday" "Thursday"
Surely this should always be Friday (except for the last point which is today, Thursday)?
What am I doing wrong here, and is there a workaround?
I wish to compute weekly changes on this data series from Friday to Friday.
..........................................REPRODUCIBLE EXAMPLE ........................
i <- as.Date((as.numeric(Sys.Date())-200):as.numeric(Sys.Date()))
x <- xts(rep(1, length(i)), order.by = i)
weekdays(index(to.weekly(x)))
# [1] "Sunday" "Sunday" "Sunday" "Sunday" "Sunday" "Sunday" "Sunday" "Sunday" "Sunday" "Sunday" "Sunday" "Sunday" "Sunday"
# [14] "Sunday" "Monday" "Monday" "Monday" "Monday" "Monday" "Monday" "Monday" "Monday" "Monday" "Monday" "Monday" "Monday"
# [27] "Monday" "Monday" "Monday" "Thursday"
y <- xts(rep(1, length(i)), order.by = as.POSIXct(i))
weekdays(index(to.weekly(y)))
# [1] "Sunday" "Sunday" "Sunday" "Sunday" "Sunday" "Sunday" "Sunday" "Sunday" "Sunday" "Sunday" "Sunday" "Sunday" "Sunday"
# [14] "Sunday" "Sunday" "Sunday" "Sunday" "Sunday" "Sunday" "Sunday" "Sunday" "Sunday" "Sunday" "Sunday" "Sunday" "Sunday"
# [27] "Sunday" "Sunday" "Sunday" "Thursday"
This is a bug in the CRAN version that has been fixed on R-Forge. Basically, the time component of the index wasn't being dropped and you ran into some funky business with timezones and daylight saving time.
Install from R-Forge to use the corrected function.
require(xts)
i <- Sys.Date()+200:0
x <- xts(rep(1, length(i)), order.by = i)
y <- xts(rep(1, length(i)), order.by = as.POSIXct(i))
all.equal(weekdays(index(to.weekly(x))), weekdays(index(to.weekly(y))))