resample_dynamics: Bootstrap from dynamics

In rossklin/bootpan: Bootstrapping for panel data

Description

Bootstrap samples by simulating a random walk with steps sampled (locally) from a procided time.table.

Usage

resample_dynamics(tt, num = nrow(unique(index(tt))), k,
  new.index.name = "resample", weight.name = NULL,
  weight.fun = resample_dynamics_default_weightfun, frequency = attr(tt,
  "frequency"), sample.from = frequency$from, resample.steps = TRUE, ...)

Arguments

tt	time.table to bootstrap from
num	number of series to resample (defaults to name number as tt)
k	number of nearest neighbours to resample from
new.index.name	column name for the new unambiguous indices (defaults to "resample", set to NULL not to include one)
weight.name	Name of new (auxiliary) column containing the weight of the sampled point
weight.fun	weight function for local resampling
frequency	list containing starting (from) and stopping (to) times as well as the lenght of the timestep (delta), defaults to using information from tt.
sample.from	initial year of tt to sample from, defaults to the year specified in frequency.
resample.steps	whether to resample relative (to current position) transitions instead of actual transitions (defaults to TRUE), use FALSE if you want the resampled series to consisit only of data points in the original data set
...	additional arguments to pass to weight.fun

Details

Note that only complete cases from tt are used. If no "new.index.name" is provided the procedure returns a data.table rather than a time.table.

rossklin/bootpan documentation built on May 27, 2019, 11:38 p.m.