ggplot2中geom_label()警告的原因及消除方法咨询

在R环境中完成数据转换、计算后，绘制Likert图与条形图的组合可视化图表时，控制台出现如下警告：

Warning message:
Removed 20 rows containing missing values or values outside the scale range (`geom_label()`). 

请问该警告是否与NA值有关？其产生的具体原因是什么？如何消除该警告？

完整代码如下：

library(tidyverse)
library(ggstats)
library(patchwork)
library(tibble)
library(tidyverse)
library(ggplot2)
library(ggstats)
likert_levels = c(
  "Very \n Dissatisfied",
  "Dissatisfied",
  "Neutral",
  "Satisfied",
  "Very \n Satisfied"
)


custom_colors = c(
  "Very \n Dissatisfied" = "#ed2e1c",
  "Dissatisfied" = "#e09c95",
  "Neutral" = "#85c1e9",
  "Satisfied" = "#7FF98B",
  "Very \n Satisfied" = "#04B431"
)

var_levels <- c(LETTERS[1:20])
n = 500
likert_levels = c(
  "Very \n Dissatisfied",
  "Dissatisfied",
  "Neutral",
  "Satisfied",
  "Very \n Satisfied"
)

df <- tibble(
  var = sample(var_levels, n, replace = TRUE),  
  val1 = sample(likert_levels, n, replace = TRUE),
  val2 = sample(c(likert_levels, NA),n, replace = TRUE),
  val3 = sample(likert_levels, n, replace = TRUE)
)

df2 = df%>%
  pivot_longer(!var, names_to = "Categories", values_to = "likert_values")%>%
  select(-Categories)%>%
  tidyr::drop_na()


df_bar = df%>%
  select(var)%>%
  group_by(var)%>%
  summarise(n=n())

df_likert = df2 %>%
  group_by(var, likert_values) %>%             # Group by `var` and `likert_values`
  summarise(count = n(), .groups = "drop") %>% # Count the occurrences
  group_by(var) %>%                            # Group by `var`
  mutate(percentage = (count / sum(count)) * 100) %>% # Calculate percentages
  ungroup()                                    # Ungroup for a clean output


df = df_likert%>%
  left_join(.,df_bar,by = "var")%>%
  select(-count)%>%
  pivot_wider(names_from = likert_values, values_from = percentage)%>%
  dplyr::relocate(var,.before=n)%>%
  dplyr::relocate(n,.before=`Very \n Dissatisfied`)%>%
  dplyr::relocate(`Very \n Dissatisfied` ,.after = n)%>%
  dplyr::relocate( Dissatisfied,.after = `Very \n Dissatisfied`)%>%
  dplyr::relocate(Neutral,.after =Dissatisfied )%>%
  dplyr::relocate(Satisfied,.after=Neutral)%>%
  dplyr::relocate(`Very \n Satisfied`,.after = Satisfied)


levels <- names(df)[-c(1:2)]
df_long <- df %>%
  select(-n) %>%
  pivot_longer(!var, names_to = "Likert", values_to = "Percentage") |>
  mutate(Likert = factor(Likert, levels))




df_tot <- df_long |>
  summarise(
    prop_lower = sum(Percentage[Likert %in% levels[1:2]]),
    prop_higher = sum(Percentage[Likert %in% levels[4:5]]),
    .by = var
  ) |>
  pivot_longer(-var,
               names_prefix = "prop_",
               values_to = "Percentage",
               names_to = "where"
  )

var_ordered <- levels(with(df_tot, reorder(var,
                                          ifelse(where == "higher", Percentage, NA),
                                          na.rm = TRUE   )) )
var_ordered = var_ordered[1:10]

df_long=df_long%>%
  filter(var %in% var_ordered)

# Likert plot
likert_plot <- ggplot(df_long, aes(x = Percentage, y = var, fill = Likert)) +
  geom_col(position = position_likert(reverse = FALSE)) +
  geom_text(
    aes(
      label = label_percent_abs(hide_below = .01, accuracy = 1, scale = 1)(Percentage)
    ),
    position = position_likert(vjust = 0.5, reverse = FALSE),
    size = 3.5,
    fontface = "bold"
  ) +
  geom_label(
    data = df_tot,
    aes(
      label = label_percent_abs(hide_below = .01, accuracy = 1, scale = 1)(Percentage),
      x = ifelse(where == "lower", -.8 , .8),
      fill = NULL
    ),
    size = 3.5,
    fontface = "bold",
    label.size = 0.2,
    show.legend = FALSE
  ) +
  scale_x_continuous(
    labels = label_percent_abs()
  ) +
  labs(
    title = "Likert Responses by Category",
    x = "Category",
    y = "Percentage",
    fill = "Likert Scale"
  ) +
  theme_bw() +
  theme( panel.border = element_rect(color = "black"))+
  scale_fill_manual(values = custom_colors) +
  labs(x = NULL, y = NULL, fill = NULL) +
  coord_cartesian(clip = "off")+
  scale_y_discrete(limits = var_ordered)



df = df%>%
  filter(var %in% var_ordered)
# Horizontal bar plot
bar_plot <- ggplot(df, aes(x = n, y = var)) +
  geom_bar(stat = "identity", fill = "lightgrey") +
  geom_label(
    aes(
      label = label_number_abs(hide_below = .05, accuracy = 2)(n)
    ),
    size = 3.5,
    position = position_stack(vjust = 0.5),
    hjust = 1,
    fill = NA,
    label.size = 0,
    color = "black"
  ) +
  scale_y_discrete(limits = var_ordered)+
  scale_x_continuous(
    labels = label_percent_abs(),
    expand = c(0, .15)
  ) +
  theme_light() +
  theme(
    legend.position = "bottom",
    panel.grid.major.y = element_blank(),
    panel.border = element_rect(color = "black") ,
    axis.text.x = element_blank() # Hides x-axis numbers
  ) +
  labs(x = NULL, y = NULL, fill = NULL)

# Print plots

(likert_plot) + (bar_plot) +
  plot_layout(
    width = c(4, 1)
  ) &
  theme(legend.position = "bottom")

1. 该警告是否与NA值有关？ #

无关，原始数据中的NA已经通过drop_na()被处理干净，本次警告和NA值没有直接关联。

2. 具体原因 #

警告来自likert_plot中的geom_label()层，核心问题是数据源不同步：

• 你通过var_ordered = var_ordered[1:10]筛选出了要展示的10个var，并对df_long、df都做了同步过滤，但geom_label()使用的数据源df_tot仍然保留了全部20个var的数据。
• 绘图时scale_y_discrete(limits = var_ordered)只显示前10个var，df_tot中剩下的10个var对应的标签数据无法在当前y轴范围内展示，被ggplot自动移除，这就是警告中“20行”的来源（每个未被筛选的var对应df_tot里lower和higher两行数据，10*2=20）。

3. 消除警告的方法 #

对df_tot同步执行筛选操作，保留和其他数据集一致的var即可：

在定义var_ordered之后，添加一行过滤代码：

df_tot = df_tot %>% filter(var %in% var_ordered)

修改后的关键代码段：

var_ordered = var_ordered[1:10]

# 同步过滤df_tot，只保留需要展示的var
df_tot = df_tot %>% filter(var %in% var_ordered)

df_long=df_long%>%
  filter(var %in% var_ordered)

这样geom_label()的数据源就只包含要显示的10个var的数据，不会再有超出y轴范围的行，警告会自动消除。

内容的提问来源于stack exchange，提问作者Homer Jay Simpson