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
|
add.ps<-function(data,col.id,col.offset,col.x,col.value,fun.aggregate="mean",
ref.offset=1,prefixes,alternative=c("two.sided","less","greater"),mu=0,
paired=FALSE,var.equal=FALSE,lty=0,...) {
round.ps<-function(x) {
as.character(ifelse(x < 0.001,"<.001",
ifelse(x < 0.01,paste0("=",substr(round(x,3),2,5)),
ifelse(x < 0.05 && round(x,2) == 0.05,"<.05",
ifelse(round(x,2) == 1,">.99",paste0("=",substr(round(x,2),2,4)))))))
}
if(!is.data.frame(data)) stop("data must be a data.frame")
columns <- c(col.id, col.offset, col.x, col.value)
if (any(!(columns %in% colnames(data)))) stop("column not matching the data")
formula.agg <- as.formula(paste(col.value, "~", col.id, "+", col.offset, "+", col.x))
d.new <- aggregate(formula.agg, data = data, FUN = fun.aggregate)
#so far same as in rm.plot2
l.offset <- levels(factor(d.new[,col.offset]))
l.offset.test <- l.offset[-ref.offset]
if (missing(prefixes)) prefixes <- paste("p(", l.offset.test, ")", sep = "")
l.x <- levels(factor(d.new[,col.x]))
for (c.offset in seq_along(l.offset.test)) {
tmp.ps <- vector("numeric", length(l.x))
for (c.x in seq_along(l.x)) {
tmpx <- d.new[d.new[,col.offset] == l.offset[ref.offset] & d.new[,col.x] == l.x[c.x], ]
tmpx <- tmpx[order(tmpx[,col.id]),]
tmpy <- d.new[d.new[,col.offset] == l.offset.test[c.offset] & d.new[,col.x] == l.x[c.x], ]
tmpy <- tmpy[order(tmpy[,col.id]),]
tmp.ps[c.x] <- t.test(tmpx[,col.value], tmpy[,col.value], alternative = alternative, mu = mu, paired = paired, var.equal = var.equal)$p.value
}
tmp.labels <- paste(prefixes[c.offset], round.ps(tmp.ps), sep = "")
axis(1,seq_along(l.x), labels = tmp.labels, line = c.offset, lty = lty, ...)
}
}
|
