如何用R的ggplot2合理创建条形图？解决温度变量显示问题

由于temperature变量的数值量级（0.6左右）远小于precipitation和radiation（100-300区间），直接绘制分组条形图会导致前者的柱子几乎不可见，以下是几种可行的优化方法：

方法1：使用双Y轴适配不同量级 #

通过设置两个Y轴，分别对应小数值的temperature和大数值的另外两个变量，同时用不同视觉元素区分：

library(tidyverse)

df <- tibble(
  type = c('grass','forest','desert','crop'),
  temperature = c(0.6206,0.6132,0.6235,0.6134),
  precipitation = c(252,290,198,337),
  radiation = c(156,177,123,205)
)

# 拆分数据集：温度和其他变量
temp_df <- df %>% select(type, temperature) %>% rename(v = temperature) %>% mutate(n = "temperature")
other_df <- df %>% select(type, precipitation, radiation) %>% pivot_longer(cols = -type, names_to = 'n', values_to = 'v')

ggplot() +
  # 绘制大数值变量的条形图（左侧Y轴）
  geom_bar(data = other_df, aes(x = type, y = v, fill = n), stat = 'identity', position = 'dodge', width = 0.7) +
  # 绘制温度的点和折线（右侧Y轴，缩放数值适配轴范围）
  geom_point(data = temp_df, aes(x = type, y = v * 500, color = n), size = 3) +
  geom_line(data = temp_df, aes(x = type, y = v * 500, color = n), group = 1) +
  # 设置双Y轴
  scale_y_continuous(
    name = "Precipitation/Radiation",
    sec.axis = sec_axis(~ . / 500, name = "Temperature")
  ) +
  scale_fill_manual(values = c("precipitation" = "#1f77b4", "radiation" = "#ff7f0e")) +
  scale_color_manual(values = c("temperature" = "#2ca02c")) +
  labs(title = "环境指标按土地类型分布", fill = "", color = "") +
  theme_minimal()

方法2：拆分变量为子图（分面展示） #

将三个变量分别绘制在独立子图中，每个子图使用自身的Y轴范围，彻底避免量级干扰：

dff <- df %>% pivot_longer(cols = -type, names_to = 'n', values_to = 'v')

ggplot(data = dff) +
  geom_bar(aes(x = type, y = v, fill = n), stat = 'identity', position = 'dodge') +
  facet_wrap(~n, scales = "free_y") +  # 每个子图使用独立Y轴
  labs(title = "环境指标按土地类型分布", x = "土地类型", y = "数值") +
  theme_minimal()

方法3：对变量做标准化处理 #

将所有变量缩放至同一范围（如Z-score标准化，均值为0、标准差为1），直观比较各变量在不同类型中的相对变化：

dff <- df %>% 
  pivot_longer(cols = -type, names_to = 'n', values_to = 'v') %>%
  group_by(n) %>%
  mutate(v_scaled = scale(v)) %>%  # 按变量分组标准化
  ungroup()

ggplot(data = dff) +
  geom_bar(aes(x = type, y = v_scaled, fill = n), stat = 'identity', position = 'dodge') +
  labs(title = "标准化环境指标按土地类型分布", x = "土地类型", y = "Z分数") +
  theme_minimal()

方法4：添加数值标签强化小数值展示 #

如果坚持使用单Y轴，给temperature的柱子添加数值标签，让观众直接读取具体数值：

dff <- df %>% pivot_longer(cols = -type, names_to = 'n', values_to = 'v')

ggplot(data = dff) +
  geom_bar(aes(x = type, y = v, fill = n), stat = 'identity', position = 'dodge') +
  geom_text(
    data = filter(dff, n == "temperature"),
    aes(x = type, y = v + 5, label = round(v, 4)),  # 标签位置略高于柱子
    color = "red", size = 3
  ) +
  labs(title = "环境指标按土地类型分布", x = "土地类型", y = "数值") +
  theme_minimal()

内容的提问来源于stack exchange，提问作者jackywang