PgaMsgl: Proximal Gradient Algorithm for Multi-variate Sparse Group...

In TriangularCell/PgaMsgl: Proximal Gradient Algorithm for Multi-variate Sparse Group Lasso

Description

Proximal Gradient Algorithm for Multi-variate Sparse Group Lasso.

Usage

PgaMsgl(XX, YY, B0, model = c("L020v1", "L020v2", "L121"), Gm, mi = 1000, mg, mc, minlambda = 1e-5, rlambda = 0.98, mintau = 1e-5, rtau = 0.98)

Arguments

XX	Matrix X in the model Y=XB.
YY	Matrix Y in the model Y=XB.
B0	Initial matrix of the coefficient matrix B in the model Y=XB.
model	The model for Sparse Group Lasso, L121, L020v1, or L020v2.
Gm	Matrix of the group structure of coefficient matrix B. It is the a matrix of group boundaries, with each row indicating a group, four columns indicate the row-start, row-end, column-start and column-end of the group. The row/column index is 1-based.
mi	Maximum number of iterations allowed, default value is 1000.
mg	An interger indicates maximum number of groups in matrix B to be reserved.
mc	An interger indicates maximum number of single coefficients in matrix B to be reserved.
minlambda	Minimum value of lambda. Only used when model = "L020v2", default value is 1e-5.
rlambda	Rate of lambda decrease. Only used when model = "L020v2", defalult value is 0.98.
mintau	Minimum value of tau. Only used when model = "L020v2", default value is 1e-5.
rtau	Rate of tau decrease. Only used when model = "L020v2", default value is 0.98.

Value

Beta	The estimated coefficient matrix B in the model Y=XB.
Rss	A vector with length mi (maximum number of iterations), which is the residual sum of squares (RSS) at each iteration.
Tau	A vector with length mi (maximum number of iterations), which is the value of parameter tau at each iteration step.
Lambda	A vector with length mi (maximum number of iterations), which is the value of parameter lambda at each iteration step.
iteration.time	An interger, which is the times of iterations in practice.
Rss_relative	A vector with length mi (maximum number of iterations), which is Rss divided by square of Y's norm.

Examples

data(lowD)
result <- PgaMsgl(lowD$X, lowD$Y, lowD$B0, model="L121", lowD$Gm, lowD$mi, lowD$mg, lowD$mc)

TriangularCell/PgaMsgl documentation built on May 28, 2019, 9:33 a.m.