Perform the local screening to "thin out" redundant break points.
second.step.local(
method = "sparse",
data,
eta,
q,
max.iteration = 1000,
tol = 10^(-4),
pts,
an,
phi.est.full = NULL,
blocks = NULL,
use.BIC = FALSE,
group.case = "columnwise",
group.index = NULL
)A list object, which contains the followings
a set of selected break point after the second local screening step
the selected Omega value
method: sparse, group sparse
input data matrix, with each column representing the time series component
tuning parameter eta for lasso
the AR order
max number of iteration for the fused lasso
tolerance for the fused lasso
the selected break points after the first step
the neighborhood size a_n
parameter matrix
a vector of blocks
use BIC for k-means part
group sparse pattern: columnwise, rowwise.
group index for group sparse case