Difference between revisions of "Locfdr.R"
From Organic Design wiki
m |
(fudge to add N(0,1) under NULL) |
||
| Line 22: | Line 22: | ||
library(locfdr) | library(locfdr) | ||
| − | locfdr(teststat, | + | locfdr(teststat, bre=100, type=1) |
| − | |||
| + | locfdr(teststat, bre=100, type=0) | ||
| + | |||
| + | a <- hist(teststat, breaks=100, plot=FALSE) | ||
q0 <- seq(-4,4, length=1001) | q0 <- seq(-4,4, length=1001) | ||
null <- dnorm(q0) | null <- dnorm(q0) | ||
| − | lines(q0, null* | + | lines(q0, null*sum(a$counts)*diff(a$breaks)[1], type="l", col="red") |
| − | + | # prob=TRUE | |
| − | + | a <- hist(teststat, breaks=120, prob=TRUE) | |
| − | + | q0 <- seq(-4,4, length=1001) | |
| − | lines( | + | null <- dnorm(q0) |
| + | lines(q0, null, type="l", col="red") | ||
Latest revision as of 12:47, 10 July 2007
Code snipits and programs written in R, S or S-PLUS
- 1) simulation page 17
p <- 1000 n <- 10 Ind <- matrix(rep(rep(c(-1,1), each=n), each=p), nr=p)
Sb <- 20
u <- matrix(rnorm(p*2*n, 0, 1 ), nr=p, nc=2*n)
B <- matrix(rnorm(p*2*n, 0, Sb), nr=p, nc=2*n)
X <- u + B*Ind/2
cl <- rep(0:1, each=n)
- 2) multtest
library(multtest) teststat <- mt.teststat(X, cl) hist(teststat)
- 3) locfdr
library(locfdr)
locfdr(teststat, bre=100, type=1)
locfdr(teststat, bre=100, type=0)
a <- hist(teststat, breaks=100, plot=FALSE) q0 <- seq(-4,4, length=1001) null <- dnorm(q0) lines(q0, null*sum(a$counts)*diff(a$breaks)[1], type="l", col="red")
- prob=TRUE
a <- hist(teststat, breaks=120, prob=TRUE) q0 <- seq(-4,4, length=1001) null <- dnorm(q0) lines(q0, null, type="l", col="red")
