gen_bank: Generate a bank of time series for use in a Near Unit...

In DataScienceForPublicPolicy/hypML: unitrootML - Machine Learning-Based Hypothesis Testing for Time Series

Description

Returns a list object of time series objects with their attributes. Whether a time series has a unit root is drawn at random based on a sampling probability (sample_prob). The length of the series is randomly drawn between t_min and t_max. The type of unit root process follows standard DGPs which can be calibrated and the split of each DGP can be controlled.

Usage

gen_bank(
  iter = 5e+05,
  sample_prob = 0.5,
  t = c(5, 50),
  freq = 12,
  nur_ur = c(0.9, 0.99999),
  run_par = TRUE,
  dgp_params = list(dgp = "dgp_enders1", periods = 100, sd = 1, alpha0 = 1, alpha2 =
    0.005, gamma = 1)
)

Arguments

iter	The number of series that will be generated.
sample_prob	The probability of that a unit root exists in the sample. For example, 0.5 would indicate 50% of series have a unit root.
t	A vector of length 2 that contains the minimum and maximum number of periods (default = c(5,50))
freq	Time series frequency (default = 12)
nur_ur	A vector of length 2 that contains the Phi domain (default = c(0.9, 0.99999))
run_par	Boolean indicating whether to compute in parallel.
dgp_params	A list object indicating which DGPs will be used and in what proportion within the sample_prob (see DGP functions). The basic DGP should include a tag "dgp" with a string value indicating which dgp function (e.g. "dgp_enders1", "dgp_engle") and the arguments required to execute the function. The parameters "period" will be handled by "t" and "gamma" is a function of the "sample_prob" parameter. Note that this function that DGPs will be evenly split amongst the sample_prob. For example, if sample_prob = 0.5 and two DGPs are specified, 25% of the overall sample will be generated using each DGP.

Value

A list object

Author(s)

Gary Cornwall and Jeffrey Chen

Examples

#Generate a new time series
gen_bank(iter = 300,
        dgp_params = list(dgp = "dgp_enders1",
                          periods = 100,
                          sd = 1,
                          alpha0 = 1,
                          alpha2 =  .005,
                          gamma = 1))

DataScienceForPublicPolicy/hypML documentation built on Dec. 17, 2021, 4:06 p.m.