bootstrap {ttrTests} | R Documentation |
Given a data set, this function returns a randomly generated data set of the same size using the bootstrap procedure on the raw data. For now, standard (i.i.d.) bootstrapping and Stationary (block) bootstrapping are supported. If random data is desired from other model distributions, for example ARIMA or GARCH, a user defined function can be input.
bootstrap(x, model = "bootstrap", userParams = 4)
x |
The data set (a univariate series) |
model |
Currently built in choices are "bootstrap" and "stationaryBootstrap". Also accepts a user defined function whose output is a series of the same length as the input data. |
userParams |
Will be passed to the function 'model', in the case that 'model' is a user defined function. Hence, a user defined function should take two parameters, the data and a list of other needed inputs. If "stationaryBootstrap" is used, userParams is the average block length from a geometric distribution, i.e. (1/lambda). |
By design the bootstrapping procedure produces samples with the same statistical properties as the original data. If a user defined function is used that generates samples with mis-matching statistical properties, these samples will not likely be useful.
sample - a univariate series the same length as the input series
A USER DEFINED MODEL MUST HAVE THE FOLLOWING FORM: function ( x , userParams ) Where x is the data, userParams is a numeric vector.
David St John
B. Efron. Bootstrap methods: Another look at the jackknife. The Annals of Statistics, 7(1):1-26, 1979.
Politis, Dimitris, and Joseph Romano, 1994, The stationary bootstrap, Journal of the American Statistical Association 89, 1303-1313.
foo <- runif(100) mean(foo) var(foo) plot(foo) sample <- bootstrap(foo) mean(sample) var(sample) plot(sample)