使用ggplot2在直方图上绘制置信区间

Question

下面的代码绘制了均值的抽样分布图，并计算了 20 批 95% 的置信区间。如何在直方图上绘制置信区间，如下面的 Photoshop 图片所示？

# plot sampling distribution of mean -----------------------------------------------------------
set.seed(1)

population <- rnorm(10000, 3, 3)

population_mean <- mean(population)

my_sample <- sample(population, 100, replace = FALSE)

standard_error <- sqrt(var(my_sample)/length(my_sample))

sampling_distribution_of_mean <- rnorm(10000, mean = population_mean, sd = standard_error)

library(ggplot2)
ggplot(data.frame(x = sampling_distribution_of_mean), aes(x)) + geom_histogram() + geom_vline(xintercept = population_mean, color = "red")


# calculate 20 lots of 95% confidence intervals -----------------------------------------------------------

my_confidence_intervals <- function(){

    my_sample <- sample(population, 100, replace = FALSE)

    sample_mean <- mean(my_sample)

    standard_error <- sqrt(var(my_sample)/length(my_sample))

    margin_of_error <- 1.96*standard_error

    mean_minus_margin_of_error <- sample_mean - margin_of_error
    mean_plus_margin_of_error <- sample_mean + margin_of_error

    c(mean_minus_margin_of_error, mean_plus_margin_of_error)

}

library(plyr)
llply(1:20, function(x) my_confidence_intervals())

Answer 1

您可能想要构建一个包含间隔的 data.frame，然后添加一层水平误差条来绘制它们。首先，我将您的范围转换为 data.frame

xx<-llply(1:20, function(x) my_confidence_intervals())
xx<-data.frame(y=1:20*50, x=do.call(rbind, xx))

现在我将它们添加到绘图中

ggplot(data.frame(x = sampling_distribution_of_mean), aes(x)) + 
    geom_histogram() + 
    geom_vline(xintercept = population_mean, color = "red") + 
    geom_errorbarh(aes(y=y, x=x.1, xmin=x.1, xmax=x.2), data=xx, col="#0094EA", size=1.2)

给出

请注意，我在创建 data.frame 时为每个范围明确设置了 y 值。

使用ggplot2在直方图上绘制置信区间

Draw confidence interval on histogram with ggplot2

r

ggplot2

statistics

data-visualization

normal-distribution