Sampling time for the first or last observation of a time series.
USAGE:
start(x, format = F, base = <<see below>>, tol = .Options$ts.eps)
end(x, format = F, base = <<see below>>, tol = .Options$ts.eps)
REQUIRED ARGUMENTS:
x:
a univariate or multivariate time series.
OPTIONAL ARGUMENTS:
format:
logical flag: if TRUE the result is given in (base, offset) format for
regular time series with integer frequency.
Default is FALSE.
base:
number which, if format is TRUE, gives the base value for the
starting time of the series from which the (base, offset) value of
start or end is to be computed
Default is floor(start.time), where start.time is the starting time of the
series.
tol:
very small number specifying the tolerance for determining whether or
not the value in question is a muliple of the base plus an integer
fraction of the frequency.
Default is .Options$ts.eps
VALUE:
start returns the time at which the first observation in x is sampled,
while end returns the time at which the last observation in x is sampled.
DETAILS:
These are generic functions with methods for each time series class.
BUGS:
For old-style time series, such as those created by ts, the assumption
is that if the frequency of the time
series is an integer, then all time values including start and end can be
represented as an integer plus some fraction that is equal to an integer
divided by the frequency. Time series created by ts have been
superseded by the classed time series created by rts, cts, and its.
You can coerce an old-style time serties to a new with as.rts.