如何去除ggplot离散变量修改版雨云图的轴侧空白区域

你遇到的问题是使用ggplot绘制含离散变量的雨云图时，large类别与轴之间存在空白，且scale_x_discrete无效。这是因为coord_flip()交换了x、y轴，原本的离散变量category对应到了y轴，而非x轴。

问题复现代码 #

# Prepare data
sample_data <- iris %>%
    filter(Species == "setosa" | Species == "virginica") %>%
    mutate(category = case_when(Sepal.Length > 5 ~ "large",
                                Sepal.Length <= 5 ~ "small")) %>%
    select(Sepal.Width,
           Species,
           category)
    
# Figure
ggplot(data = sample_data) +
    aes(x = category,
        y = Sepal.Width,
        fill = category) +
    ggdist::stat_halfeye(
        adjust = 0.5,
        justification = -.1,
        .width = 0,
        point_colour = NA
    ) +
    geom_boxplot(width = .05,
                 outlier.colour = NA,
                 alpha = 0.5) +
    ggdist::stat_dots(
        side = "left",
        justification = 1.1,
        binwidth = 0.015) +
    facet_grid(~ Species) +
    labs(title = "Figure 1: Change in sepal width with sepal length",
         x = "Sepal length") +
    coord_flip() +
    tidyquant::scale_fill_tq() +
    tidyquant::theme_tq()

问题图示 #

解决方案 #

核心是调整y轴（原x轴）的离散刻度扩展参数，同时微调图形元素的位置避免截断：

1. 使用scale_y_discrete(expand = c(0, 0.05))消除离散轴的默认空白，0.05是为了保留少量空隙，防止图形元素被轴截断；
2. 调整stat_halfeye和stat_dots的justification参数，让图形元素更贴近轴。

修改后的完整代码：

# Prepare data
sample_data <- iris %>%
    filter(Species == "setosa" | Species == "virginica") %>%
    mutate(category = case_when(Sepal.Length > 5 ~ "large",
                                Sepal.Length <= 5 ~ "small")) %>%
    select(Sepal.Width,
           Species,
           category)
    
# Figure
ggplot(data = sample_data) +
    aes(x = category,
        y = Sepal.Width,
        fill = category) +
    ggdist::stat_halfeye(
        adjust = 0.5,
        justification = 0,  # 调整为0让半眼图贴近轴
        .width = 0,
        point_colour = NA
    ) +
    geom_boxplot(width = .05,
                 outlier.colour = NA,
                 alpha = 0.5) +
    ggdist::stat_dots(
        side = "left",
        justification = 1.0,  # 同步调整点的位置
        binwidth = 0.015) +
    facet_grid(~ Species) +
    labs(title = "Figure 1: Change in sepal width with sepal length",
         x = "Sepal length") +
    coord_flip() +
    scale_y_discrete(expand = c(0, 0.05)) +  # 消除离散轴空白
    tidyquant::scale_fill_tq() +
    tidyquant::theme_tq()

原理说明 #

coord_flip()会交换x和y轴的映射关系，原本的离散变量category从x轴转移到了y轴。之前使用scale_x_discrete修改的是翻转后的数值轴（原y轴），自然无法消除离散轴的空白。通过scale_y_discrete调整离散轴的expand参数，就能去掉默认的空白padding。

内容的提问来源于stack exchange，提问作者Tory