stage.forward: Nonlinear Forward stagewise regression using DCOL

In nlnet: Nonlinear Network, Clustering, and Variable Selection Based on DCOL

Description

The subroutine conducts forward stagewise regression using DCOL. Either DCOL roughening or spline roughening is conducted.

Usage

stage.forward(X, y, step.size = 0.01, stop.alpha = 0.01, 
stop.var.count = 20, roughening.method = "DCOL", tol = 1e-08, 
spline.df = 5, dcol.sel.only = FALSE, do.plot = F)

Arguments

X	The predictor matrix. Each row is a gene (predictor), each column is a sample. Notice the dimensionality is different than most other packages, where each column is a predictor. This is to conform to other functions in this package that handles gene expression type of data.
y	The numerical outcome vector.
step.size	The step size of the roughening process.
stop.alpha	The alpha level (significance of the current selected predictor) to stop the iterations.
stop.var.count	The maximum number of predictors to select. Once this number is reached, the iteration stops.
roughening.method	The method for roughening. The choices are "DCOL" or "spline".
tol	The tolerance level of sum of squared changes in the residuals.
spline.df	The degree of freedom for the spline.
dcol.sel.only	TRUE or FALSE. If FALSE, the selection of predictors will consider both linear and nonlinear association significance.
do.plot	Whether to plot the points change in each step.

Details

Please refer to the reference manuscript for details.

Value

A list object is returned. The components include the following.

found.pred	The selected predictors (row number).
ssx.rec	The magnitude of variance explained using the current predictor at each step.
$sel.rec	The selected predictor at each step.
$p.rec	The p-value of the association between the current residual and the selected predictor at each step.

Author(s)

Tianwei Yu<[e-mail]>

References

https://arxiv.org/abs/1601.05285

See Also

nvsd

Examples

X<-matrix(rnorm(2000),ncol=20)
y<-sin(X[,1])+X[,2]^2+X[,3]
stage.forward(t(X),y,stop.alpha=0.001,step.size=0.05)

nlnet documentation built on Jan. 13, 2021, 10:35 a.m.