rdataframeggplot2dplyr

Warning message in ggplot2 `geom_label()`


I have a data frame in R that after some data transforming and calculations and plotting :

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")

I receive the :

enter image description here

but in the console I receive a warning message :

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

why I receive this warning ? is something regarding the NA's ? How can I stop this ?


Solution

  • The issue is that df_tot includes all categories of var. But as you set limits=var_ordered all categories not in var_ordered get dropped as they fall outside of the limits, i.e. these categories are values outside the scale range.

    To silent the warning you can use data = df_tot |> filter(var %in% var_ordered) in geom_label to include only the categories in var_ordered.

    library(tidyverse)
    library(ggstats)
    library(patchwork)
    
    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 |> filter(var %in% var_ordered),
        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)
    

    ...

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