在R中绘制两个含不同着色区间的正态分布图形

方案一：使用ggplot2的stat_function（推荐，无需预生成数据） #

直接用stat_function调用正态分布密度函数，通过xlim参数精准控制每个分布的着色区间，避免数据框拼接带来的颜色混乱问题。

完整可复现代码 #

library(ggplot2)

# 定义核心参数
cutoff <- 2          # 临界值
h0_mean <- 0         # H0分布均值
h0_sd <- 1           # H0分布标准差
h1_mean <- 2.5       # H1分布均值
h1_sd <- 1           # H1分布标准差

# 绘制图形
ggplot(data.frame(x = c(-3, 5)), aes(x = x)) +
  # 绘制H0分布：x≥2标红，其余灰色
  stat_function(fun = dnorm, args = list(mean = h0_mean, sd = h0_sd),
                geom = "area", fill = "gray70", xlim = c(-3, cutoff)) +
  stat_function(fun = dnorm, args = list(mean = h0_mean, sd = h0_sd),
                geom = "area", fill = "red", alpha = 0.5, xlim = c(cutoff, 5)) +
  # 绘制H1分布：x≤2标蓝，其余灰色
  stat_function(fun = dnorm, args = list(mean = h1_mean, sd = h1_sd),
                geom = "area", fill = "gray70", xlim = c(cutoff, 5)) +
  stat_function(fun = dnorm, args = list(mean = h1_mean, sd = h1_sd),
                geom = "area", fill = "blue", alpha = 0.5, xlim = c(-3, cutoff)) +
  # 添加临界虚线
  geom_vline(xintercept = cutoff, linetype = "dashed", color = "black") +
  # 标注分布名称
  annotate("text", x = h0_mean, y = dnorm(h0_mean, h0_mean, h0_sd),
           label = "H0", vjust = -0.5) +
  annotate("text", x = h1_mean, y = dnorm(h1_mean, h1_mean, h1_sd),
           label = "H1", vjust = -0.5) +
  # 调整标签和主题
  labs(x = "X", y = "Density", title = "G*Power Style Distribution Plot") +
  theme_minimal()

方案二：基于预生成数据框精准着色 #

如果坚持使用预生成数据，可通过为每个数据点指定颜色值，再用scale_fill_identity直接映射颜色，解决重叠区着色错误问题。

完整可复现代码 #

library(dplyr)
library(ggplot2)

# 生成H0分布数据并指定颜色
h0_data <- data.frame(x = seq(-3, 5, length.out = 1000)) %>%
  mutate(
    y = dnorm(x, mean = 0, sd = 1),
    dist = "H0",
    fill_color = ifelse(x >= 2, "red", "gray70")
  )

# 生成H1分布数据并指定颜色
h1_data <- data.frame(x = seq(-3, 5, length.out = 1000)) %>%
  mutate(
    y = dnorm(x, mean = 2.5, sd = 1),
    dist = "H1",
    fill_color = ifelse(x <= 2, "blue", "gray70")
  )

# 合并数据
both_data <- bind_rows(h0_data, h1_data)

# 绘图
ggplot(both_data, aes(x = x, y = y, fill = fill_color)) +
  geom_area(alpha = 0.5) +
  geom_vline(xintercept = 2, linetype = "dashed", color = "black") +
  scale_fill_identity() +  # 直接使用fill_color列的颜色值
  labs(x = "X", y = "Density", title = "G*Power Style Distribution Plot") +
  theme_minimal()

问题原因说明 #

你之前的代码出错，是因为合并数据框后，着色逻辑没有针对每个分布单独过滤x范围，导致重叠区域的颜色被错误覆盖。上面两种方案分别通过分段绘制函数、为每个数据点绑定专属颜色，精准实现了H0(x≥2红)、H1(x≤2蓝)的着色需求。

内容的提问来源于stack exchange，提问作者wes