expr.dev.test: Deviance Test

Description Usage Arguments Value Author(s) See Also Examples

View source: R/expr.dev.test.R

Description

Deviance permuation test.

Usage

1
2
3
4
5
6
7
expr.dev.test(xx, formula.full, formula.red = NULL, D.red = NULL, model.dat, 
	test.vars, glm.family, perm = 100, method = c("chisqstat", "permutation"), 
	cf="fisher", adjust=FALSE, snowfall.args=list(parallel=FALSE), 
	snowfall.seed, save.stat.data = FALSE, 
	file.save = "GlobalDeviance_statistic_Tperm_and_Tobs.Rdata",
	use.save.stat.data = FALSE, 
	file.use = "GlobalDeviance_statistic_Tperm_and_Tobs.Rdata")

Arguments

xx

Dataset (variables x observations/patients).

formula.full

Formula from the full model.

formula.red

Formula from the reduced model.

D.red

Designmatrix of the reduced model.

model.dat

Dataset with covariables (observations/patients x covariables).

test.vars

Variables or groups of variables to test.

glm.family

Family for the regression.

perm

Number of permutations.

method

Method for the calculation of the p-value.

cf

Combining function for the partial tests.

adjust

Method for p-value adjustment. Not implemented.

snowfall.args

A list with parameters for the initialisation of parallelizing with package snowfall.

snowfall.seed

Start seed.

save.stat.data

Logical value, default is FALSE. Save the permutation and original test statistic.

file.save

File name for saving.

use.save.stat.data

Logical value, default is FALSE. Use the saved permutation and original test statistic.

file.use

File name for data loading.

Value

A list

method

Method for the calculation of the p-value.

number.of.variables

Number of variables.

number.of.permutations

Number of permutations.

formula.full

Formula from the full model.

formula.red

Formula from the reduced model.

test

Test results.

data.perm

Permutation test statistic.

data.original

Original test statistic.

test.vars

Variables or groups of variables to test.

Author(s)

Frederike Fuhlbrueck f.fuhlbrueck@googlemail.com

See Also

See PermTest, varwiselogLik and Rossi.

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
## Not run: 
### Example 1: poisson random data 
set.seed(6666)
n<-100
Y1<-c(rpois(n, 1))
Y2<-c(rpois(n/2, 1), rpois(n/2, 10))
A<-rnorm(n)
B<-c(rep(1, n/2), rep(0, n/2))       # group variable
C<-rpois(n, 1)

test.variables<-list("Y1", "Y2", c("Y1", "Y2"))
names(test.variables)<-c("Y1", "Y2", "Y1, Y2")

t.random<-expr.dev.test(xx=t(data.frame(Y1, Y2)), formula.full=~ A + B + C, 
	formula.red=~ A + C, model.dat=data.frame(A, B, C), test.vars=test.variables, 
	glm.family=poisson(link="log"), perm=1000, method="permutation", cf="fisher", 
	snowfall.args=list(parallel=TRUE), snowfall.seed=54321)

summary(t.random, digits=3)



### Example 2:  data set Rossi
data(Rossi)

# Covariables (patients x covariables)
model.dat<-Rossi[, c("arrest", "fin", "wexp")]
str(model.dat)

# data (variables/genes x patients)
xx<-rbind(t(t(t(Rossi[, c("prio", "n.work.weeks")]))), rpois(432, 1))
rownames(xx)<-c("prio", "n.work.weeks", "random")

formula.full<- ~ arrest + fin + wexp
formula.red<- ~ arrest + fin

test.vars<-list("prio", "n.work.weeks", "random", c("prio", "n.work.weeks"), 
	c("prio", "n.work.weeks", "random"))
names(test.vars)<-c("prio", "n.work.weeks", "random", "prio+n.work.weeks", 
	"prio+n.work.weeks+random")

set.seed(54321)

t.rossi1<-expr.dev.test(xx=xx, formula.full=formula.full, formula.red=formula.red, 
	D.red=NULL, model.dat, test.vars=test.vars, glm.family=poisson(link="log"), 
	perm=100, method="permutation", cf="fisher")

t.rossi2<-expr.dev.test(xx=xx, formula.full=formula.full, formula.red=formula.red, 
	D.red=NULL, model.dat, test.vars=test.vars, glm.family=poisson(link="log"), 
	perm=100, method="chisqstat", cf="fisher")

summary(t.rossi1, digits=2)

summary(t.rossi2, digits=3)


## End(Not run)

GlobalDeviance documentation built on May 2, 2019, 11:32 a.m.