使用链接中的数据,您可以像这样计算平均值、se 和 ci (我建议使用dplyr
, 的继承者plyr
):
set.seed(1234)
dat <- data.frame(cond = factor(rep(c("A","B"), each=200)),
rating = c(rnorm(200),rnorm(200, mean=.8)))
library(ggplot2)
library(dplyr)
cdat <- dat %>%
group_by(cond) %>%
summarise(rating.mean = mean(rating),
sem = sd(rating)/sqrt(length(rating)),
ci.low = mean(rating) - 2*sem,
ci.upp = mean(rating) + 2*sem)
为了将线条限制在曲线轮廓上,您首先需要找出曲线的哪些 y 值对应于平均值,例如通过访问密度值ggplot_build
并使用approx
:
cdat.dens <- ggplot_build(ggplot(dat, aes(x=rating, colour=cond)) + geom_density())$data[[1]] %>%
mutate(cond = ifelse(group == 1, "A", "B")) %>%
left_join(cdat) %>%
select(y, x, cond, rating.mean, sem, ci.low, ci.upp) %>%
group_by(cond) %>%
mutate(dens.mean = approx(x, y, xout = rating.mean)[[2]],
dens.cilow = approx(x, y, xout = ci.low)[[2]],
dens.ciupp = approx(x, y, xout = ci.upp)[[2]]) %>%
select(-y, -x) %>%
slice(1)
> cdat.dens
Source: local data frame [2 x 8]
Groups: cond [2]
cond rating.mean sem ci.low ci.upp dens.mean dens.cilow dens.ciupp
<chr> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
1 A -0.05775928 0.07217200 -0.2021033 0.08658471 0.3865929 0.403623 0.3643583
2 B 0.87324927 0.07120697 0.7308353 1.01566320 0.3979347 0.381683 0.4096153
然后您可以将所有内容与各种组合geom_segment
s:
ggplot() +
geom_density(data = dat, aes(x = rating, colour = cond)) +
geom_segment(data = cdat.dens, aes(x = rating.mean, xend = rating.mean, y = 0, yend = dens.mean, colour = cond),
linetype = "dashed", size = 1) +
geom_segment(data = cdat.dens, aes(x = ci.low, xend = ci.low, y = 0, yend = dens.cilow, colour = cond),
linetype = "dotted", size = 1) +
geom_segment(data = cdat.dens, aes(x = ci.upp, xend = ci.upp, y = 0, yend = dens.ciupp, colour = cond),
linetype = "dotted", size = 1)
正如 Axeman 指出的那样,您可以根据功能区区域创建多边形,如中所述这个答案 https://stackoverflow.com/questions/12429333/how-to-shade-a-region-under-a-curve-using-ggplot2.
因此,对于您的数据,您可以子集并添加附加行,如下所示:
ribbon <- ggplot_build(ggplot(dat, aes(x=rating, colour=cond)) + geom_density())$data[[1]] %>%
mutate(cond = ifelse(group == 1, "A", "B")) %>%
left_join(cdat.dens) %>%
group_by(cond) %>%
filter(x >= ci.low & x <= ci.upp) %>%
select(cond, x, y)
ribbon <- rbind(data.frame(cond = c("A", "B"), x = c(-0.2021033, 0.7308353), y = c(0, 0)),
as.data.frame(ribbon),
data.frame(cond = c("A", "B"), x = c(0.08658471, 1.01566320), y = c(0, 0)))
And add geom_polygon
到情节:
ggplot() +
geom_polygon(data = ribbon, aes(x = x, y = y, fill = cond), alpha = .5) +
geom_density(data = dat, aes(x = rating, colour = cond)) +
geom_segment(data = cdat.dens, aes(x = rating.mean, xend = rating.mean, y = 0, yend = dens.mean, colour = cond),
linetype = "dashed", size = 1) +
geom_segment(data = cdat.dens, aes(x = ci.low, xend = ci.low, y = 0, yend = dens.cilow, colour = cond),
linetype = "dotted", size = 1) +
geom_segment(data = cdat.dens, aes(x = ci.upp, xend = ci.upp, y = 0, yend = dens.ciupp, colour = cond),
linetype = "dotted", size = 1)
这是针对您的真实数据的改编代码。合并两个组而不是一个组有点棘手:
cdat <- dat %>%
group_by(direction, cond) %>%
summarise(rating.mean = mean(rating, na.rm = TRUE),
sem = sd(rating, na.rm = TRUE)/sqrt(length(rating)),
ci.low = mean(rating, na.rm = TRUE) - 2*sem,
ci.upp = mean(rating, na.rm = TRUE) + 2*sem)
cdat.dens <- ggplot_build(ggplot(dat, aes(x=rating, colour=interaction(direction, cond))) + geom_density())$data[[1]] %>%
mutate(cond = ifelse((group == 1 | group == 2 | group == 3 | group == 4), "A",
ifelse((group == 5 | group == 6 | group == 7 | group == 8), "B", "C")),
direction = ifelse((group == 1 | group == 5 | group == 9), "EAST",
ifelse((group == 2 | group == 6 | group == 10), "NORTH",
ifelse((group == 3 | group == 7 | group == 11), "SOUTH", "WEST")))) %>%
left_join(cdat) %>%
select(y, x, cond, direction, rating.mean, sem, ci.low, ci.upp) %>%
group_by(cond, direction) %>%
mutate(dens.mean = approx(x, y, xout = rating.mean)[[2]],
dens.cilow = approx(x, y, xout = ci.low)[[2]],
dens.ciupp = approx(x, y, xout = ci.upp)[[2]]) %>%
select(-y, -x) %>%
slice(1)
ggplot() +
geom_density(data = dat, aes(x = rating, colour = cond)) +
geom_segment(data = cdat.dens, aes(x = rating.mean, xend = rating.mean, y = 0, yend = dens.mean, colour = cond),
linetype = "dashed", size = 1) +
geom_segment(data = cdat.dens, aes(x = ci.low, xend = ci.low, y = 0, yend = dens.cilow, colour = cond),
linetype = "dotted", size = 1) +
geom_segment(data = cdat.dens, aes(x = ci.upp, xend = ci.upp, y = 0, yend = dens.ciupp, colour = cond),
linetype = "dotted", size = 1) +
facet_wrap(~direction)