ggplot分组直方图X轴高值区域条形不可见的解决方法

你的问题核心是数据集中在X轴低值区间，高值区间数据稀疏，导致默认等宽分箱下高值区条形被过度压缩。以下是几个贴合目标图效果的优化方案：

方案1：X轴对数变换 #

通过对数变换压缩高值区间的刻度间距，既保留整体分布，又让高值区条形清晰可见，适配偏态分布的数据：

library(ggplot2)
library(dplyr)

graph2 |>
  filter(!is.na(healthy)) |>
  ggplot(aes(x = total_visits + 1, fill = as.factor(healthy))) +  # +1避免0值对数计算报错
  geom_histogram(aes(y = after_stat(count / sum(count))),
                 alpha = 0.6, color = "white", position = 'identity',
                 breaks = exp(seq(log(1), log(101), length.out = 20))) +  # 对数间距分箱
  scale_x_log10(breaks = c(1, 5, 10, 20, 30, 50, 100),
                labels = c(0, 5, 10, 20, 30, 50, 100)) +  # 标签映射回原始数值
  scale_fill_manual(labels = c("TSCI", "SHS"), values = c("blue", "red")) +
  labs(fill = "", x = "total_visits")

方案2：截断X轴+嵌入高值区放大图 #

保留主图聚焦数据密集区，同时嵌入小图单独展示高值细节，兼顾整体与局部：

library(ggplot2)
library(dplyr)
library(ggpmisc)

# 主图：展示0-30的核心分布
main_plot <- graph2 |>
  filter(!is.na(healthy)) |>
  ggplot(aes(x = total_visits, fill = as.factor(healthy))) +
  geom_histogram(aes(y = after_stat(count / sum(count))),
                 alpha = 0.6, color = "white", position = 'identity',
                 breaks = seq(0, 30, by = 1)) +
  scale_x_continuous(breaks = seq(0, 30, 5), limits = c(0, 30)) +
  scale_fill_manual(labels = c("TSCI", "SHS"), values = c("blue", "red")) +
  labs(fill = "", x = "total_visits") +
  theme(plot.margin = margin(5, 20, 5, 5))

# 子图：放大30-100的高值区
inset_plot <- graph2 |>
  filter(!is.na(healthy), total_visits > 30) |>
  ggplot(aes(x = total_visits, fill = as.factor(healthy))) +
  geom_histogram(aes(y = after_stat(count / sum(count))),
                 alpha = 0.6, color = "white", position = 'identity',
                 breaks = seq(30, 100, by = 5)) +
  scale_x_continuous(breaks = seq(30, 100, 10)) +
  scale_fill_manual(labels = c("TSCI", "SHS"), values = c("blue", "red")) +
  labs(x = "", y = "") +
  theme_minimal() +
  theme(legend.position = "none",
        plot.background = element_rect(fill = "white", color = "black"))

# 组合主图与子图
main_plot +
  annotation_custom(grob = ggplotGrob(inset_plot),
                    xmin = 20, xmax = 30, ymin = 0.08, ymax = 0.15)

方案3：自定义分箱宽度 #

对低值区用窄分箱保留细节，高值区用宽分箱避免条形过窄，平衡不同区间的显示效果：

library(ggplot2)
library(dplyr)

# 自定义分箱断点：0-20每1单位一箱，20-100每5单位一箱
custom_breaks <- c(seq(0, 20, 1), seq(25, 100, 5))

graph2 |>
  filter(!is.na(healthy)) |>
  ggplot(aes(x = total_visits, fill = as.factor(healthy))) +
  geom_histogram(aes(y = after_stat(count / sum(count))),
                 alpha = 0.6, color = "white", position = 'identity',
                 breaks = custom_breaks) +
  scale_x_continuous(breaks = c(seq(0, 20, 5), seq(25, 100, 10))) +
  scale_fill_manual(labels = c("TSCI", "SHS"), values = c("blue", "red")) +
  labs(fill = "")

内容的提问来源于stack exchange，提问作者user3483060