time(x) cycle(x)
NAs) are allowed.
time function returns the time value of each observation.
For instance,
time(x)[1]
is the same as
start(x), the starting time of
x, and
time(x)[length(time(x))]
is the same as
end(x), the ending time of
x.
If
x is a regular time series (of class
"rts"),
time returns a
univariate time series with the same time parameters (
tspar) and
ending time as
x;
if
x is a calendar time series,
time returns a vector of dates,
and if
x is an irregular time series,
time returns the same vector
as
tspar.
cycle function returns the position of
each observation within the cycle of length
frequency
associated with the series.
If
x were monthly, this would be the month number:
1 for each January, etc.
These are generic functions with methods for each time series class.
Neither
time.its nor
time.cts return a time series since they may return
a vector of class
dates which is not an acceptable time series in the
current implementation.
The method
cycle.cts implicitly assumes a yearly cycle with one exception:
If
units="days" and
frequency=7 then it uses days of the week to
decide where the cycle begins (thereby implicitly assuming a weekly
cycle).
The
cycle function is not defined for irregular time series.
growth <- rts(sample(100:300,25), freq = 12, units = "months") fgrowth <- factor(cycle(growth), levels = 1:12, labels = month.name) tapply(growth, fgrowth, mean) # mean growth for each month x <- rts(rnorm(100), start = c(1953, 4), frequency = 12) plot(time(x), x) # similar to ts.plot(x)