full.select.gen: Perform MPS

In nkissel/ModelPath: Model Path Selection

Description

Function that performs Model Path Selection (MPS)

Usage

full.select.gen(
  myframe,
  resp.name,
  depth,
  r,
  f,
  type,
  model,
  fun.args,
  pred.args,
  condense = FALSE
)

Arguments

myframe	data.frame holding explanatory and response data
resp.name	matrix of MPS data
depth	maximum depth of paths
r	maximum cell count
f	loss function
type	'reg' or 'class' for regression and classification. Not needed if f is specified.
model	name of modeling method; must be name of the function called in R
fun.args	list of arguments that are passed to modeling function
pred.args	list of arguments that are passed to model predict function
condense	logical for whether or not paths should be merged

Examples

set.seed(200)
library(MASS)
n<-1000
p<-10
x<-mvrnorm(n,rep(0,p),diag(p))
signal<-rowSums(x[,1:3])
noise<-rnorm(n,0,1/4)
y<-signal+noise
mydata<-data.frame(x,y)
mps1<-full.select.gen(myframe=mydata,resp.name='y',depth=3,
   B=100,model='lm',condense=FALSE)
mps1
build.tree(mps1) #graph

#merged paths
mps2<-full.select.gen(myframe=mydata,resp.name='y',depth=3,
   r=100,model='lm',condense=TRUE)
mps2
build.tree(mps2) #graph

nkissel/ModelPath documentation built on Feb. 13, 2021, 12:53 a.m.