lag_selection: Select the lag of the VAR model using total BIC method

In VARDetect: Multiple Change Point Detection in Structural VAR Models

Select the lag of the VAR model using total BIC method

Description

Select the lag of the VAR model (if the lag is unknown) using BIC method for total segments

Usage

lag_selection(
  data,
  method = c("sparse", "group sparse", "fLS"),
  group.case = c("columnwise", "rowwise"),
  group.index = NULL,
  lambda.1.cv = NULL,
  lambda.2.cv = NULL,
  mu = NULL,
  block.size = NULL,
  blocks = NULL,
  use.BIC = TRUE,
  an.grid = NULL,
  threshold = NULL,
  lag_candidates,
  verbose = FALSE
)

Arguments

data	input data matrix, each column represents the time series component
method	method is sparse, group sparse and fixed lowrank plus sparse
group.case	two different types of group sparse, column-wise and row-wise, respectively.
group.index	specify group sparse index. Default is NULL.
lambda.1.cv	tuning parameter lambda_1 for fused lasso
lambda.2.cv	tuning parameter lambda_2 for fused lasso
mu	tuning parameter for low rank component, only available when method is set to "fLS".
block.size	the block size
blocks	the blocks
use.BIC	use BIC for k-means part
an.grid	a vector of an for grid searching.
threshold	a numeric argument, give the threshold for estimated model parameter matrices. Default is NULL.
lag_candidates	potential lag selection set
verbose	A Boolean argument, if TRUE, it provides detailed information. Default is FALSE

Value

selected lag for VAR series

select_lag

An integer no less than 1 represents the selected lag of time series.

Examples

nob <- 1000; p <- 15
brk <- c(floor(nob / 2), nob + 1)
m <- length(brk)
q.t <- 2 # the lag of VAR model for simulation
signals <- c(-0.8, 0.6, 0.4)
try <- simu_var(method = "sparse", nob = nob, k = p, brk = brk,
                signals = signals, lags_vector = c(1, 2),
                sp_pattern = "off-diagonal")
data <- try$series; data <- as.matrix(data)

# Apply lag selection to determine the lag for the given time series
lag_candi <- c(1, 2, 3, 4)
select_lag <- lag_selection(data = data,
                            method = "sparse", lag_candidates = lag_candi)
print(select_lag)

VARDetect documentation built on June 22, 2024, 10 a.m.