get_B_ADMM: The main function of mSIM package

In echi/mSIM: Multivariate Response Single Index Model

Description

The main function of mSIM package

Usage

get_B_ADMM(
  Y,
  X,
  B = NULL,
  lambda,
  rank,
  alpha = 1,
  control1 = list(max.iter = 100, tol = 0.01),
  control2 = list(ele.sparse = FALSE, row.sparse = TRUE, low.rank = TRUE),
  select.method = "linear",
  descent.method = "bfgs",
  plot = FALSE
)

Arguments

Y	The index matrix
X	The covariate matrix
B	The initial coefficient matrix
lambda	The regualrize coefficient
rank	The rank of B matrix
alpha	The step in the optimization process
control1	A list that contains parameters for minimize loss function
control2	A list that contains parameters for regularize
select.method	link funtion selection method
plot	whether or not plot the error during the iteration
descend.method	optimization method

Value

A list that contains information such as the final B matrix, the training error.

Examples

X <- scale(X_train)
Y <- scale(Y_train)
test = get_B_ADMM(Y=Y, X=X, lambda=0.2, rank=3, alpha=1, control1 = list(max.iter=5e1, tol=1e-5),
control2=list(ele.sparse=FALSE, row.sparse=TRUE, low.rank=TRUE), select.method='linear', plot=FALSE)

echi/mSIM documentation built on Oct. 6, 2020, 11:09 a.m.