Plotting labels with ggplot2

Packages

• Load packages using the pacman::p_load() function.

• Packages: tidyverse, here, patchwork, medicaldata.

# Load required packages:tidyverse, here, patchwork, medicaldata
pacman :: p_load(tidyverse, here, patchwork, medicaldata,knitr,DT,usethis,ggtext)

Introduction to text geoms in

• Focus on geom_text() for simple labeling and geom_label() for emphasized labels.

• Apply these geoms on bar plots, stacked bars, dodged bars, normalized stacked bars and circular plots.

• Start with a simple bar plot using fake data for practice.

# Create example data frame with categories A,B,C and counts 10,20,15
data <- 
  data.frame(
    category = c("A","B","C"),
    count = c(10, 20, 15)
  )
data

  category count
1        A    10
2        B    20
3        C    15

# Generate a bar plot using ggplot2, mapping category to x and count to y,
# Use geom_col() and fill with "steelblue"
# and geom_text () to add labels

ggplot(data, mapping = aes(
  x = category,
  y = count
)) +
  geom_col(fill = "steelblue")+
  geom_text(aes(label = count))

• As you can see however, the placement of our text is odd. Can fix this with nudge_y. Also increase text size.

ggplot(data, mapping = aes(
  x = category,
  y = count
)) +
  geom_col(fill = "steelblue")+
  geom_text(aes(label = count), nudge_y = 2, size = 5 )

• Explore negative values of nudge_y to move text down.

ggplot(data, mapping = aes(
  x = category,
  y = count
)) +
  geom_col(fill = "steelblue")+
  geom_text(aes(label = count), nudge_y = -2, size = 5 )

• For horizontal bar plots, use nudge_x to adjust horizontally.

ggplot(data, mapping = aes(
  x = count,
  y = category
)) +
  geom_col(fill = "steelblue")+
  geom_text(aes(label = count), nudge_x = 1, size = 5 )

• We can also use the geom_label () function for text labels with background rectangles.

# Replace geom_text with geom_label for labels with background
ggplot(data, mapping = aes(
  x = count,
  y = category
)) +
  geom_col(fill = "steelblue")+
  geom_label(aes(label = count), nudge_x = 1, size = 5 )

ggplot(data, mapping = aes(
  x = category,
  y = count
)) +
  geom_col(fill = "steelblue")+
  geom_label(aes(label = count), nudge_y = -1, size = 5 )

• We can fill and color aesthetics in geom_label () for background and text color.

ggplot(data, mapping = aes(
  x = category,
  y = count
)) +
  geom_col(fill = "steelblue")+
  geom_label(aes(label = count), 
             nudge_y = -1, 
             size = 5,
             color = "darkblue",
             fill = "lightblue")

# Adjust fill and color in geom_label for background and text color change

• PRO-TIP: Setting a custom theme to streamline your plotting process

• Combine an existing theme with your own customization: create a custom theme combining theme_dark with large bold axis labels:

# Define a custom theme with theme_light and large bold axis titles
theme_dark_custom <- 
  theme_dark() +
  theme(axis.title = element_text(size = 16, face = "bold"))

Can use this on a single plot as follows

ggplot(data, mapping = aes(
  x = category,
  y = count
)) +
  geom_col(fill = "steelblue")+
  geom_label(aes(label = count), 
             nudge_y = -1, 
             size = 5,
             color = "darkblue",
             fill = "lightblue")+
  theme_dark_custom

• Set the custom theme as the default for all plots

# set the custom theme as the default for all plots
theme_set(theme_dark_custom)

• Observe the effect: theme_dark_custom automatically applies to every new plot.

# Redraw an earlier plot using the new global theme
ggplot(data, aes(
  x = category,
  y = count
))+
  geom_col(fill = "steelblue")+
  geom_label(aes(label = count),
             nudge_y = -3,
             fill = "royalblue4",
             color = "white")

To set the default theme back to the original, use theme_set(theme_gray()).

theme_set(theme_gray())

ggplot(data, aes(
  x = category,
  y = count
))+
  geom_col(fill = "steelblue")+
  geom_label(aes(label = count),
             nudge_y = -3,
             fill = "royalblue4",
             color = "white")

The vjust and hjust arguments

• Explore vjust and hjust in ggplot2 for adjusting text position.

• These arguments control Vertical and horizontal justification of text.

Understanding hjust (horizontal justification)

• hjust modifies horizontal position of text relative to anchor points.

hjust values ranges from 0 to 1

• ‘hjust = 0’ : Left edge aligns with anchor point.

• ‘hjust = 0.5’ : Centered on anchor point.

• ‘hjust = 1’ : Right edge aligns with anchor point.

# Example to show hjust in action
# Use geom_text to display different hjust values ona plot

#example

df <- 
  data.frame(x=1, y=1)
# Base plot with a point

base_p <- 
  ggplot(df, aes(x,y)) +
  geom_point() +
  theme_void()
base_p + geom_text(aes(label = "text"))

p_hjust_0 <- base_p + geom_text(aes(label = "hjust =0"),hjust = 0)
p_hjust_0.25 <- base_p + geom_text(aes(label = "hjust =0.25"),hjust = 0.25)
p_hjust_0.5 <- base_p + geom_text(aes(label = "hjust =0.5"),hjust = 0.5)
p_hjust_0.75 <- base_p + geom_text(aes(label = "hjust =0.75"),hjust = 0.75)
p_hjust_1 <- base_p + geom_text(aes(label = "hjust =1"),hjust = 1)

# Combine plots with patchwork
p_hjust_0 / p_hjust_0.25 / p_hjust_0.5 / p_hjust_0.75 / p_hjust_1

# Import data from csv
tb_outcomes <- 
  read.csv("C:/Users/Perminus Njiru/OneDrive - LVCT Health/Desktop/Freecodecamp/R/Care_and_Treatment_Analysis/Input/benin_tb.csv")
head(tb_outcomes)

  period period_date        hospital     outcome cases  diagnosis_type
1 2015Q4  2015-10-01 St Jean De Dieu      failed     0 bacteriological
2 2015Q4  2015-10-01 St Jean De Dieu unevaluated     0 bacteriological
3 2015Q4  2015-10-01 St Jean De Dieu        died     0 bacteriological
4 2015Q4  2015-10-01 St Jean De Dieu        lost     0 bacteriological
5 2015Q4  2015-10-01 St Jean De Dieu   completed     0 bacteriological
6 2015Q4  2015-10-01 St Jean De Dieu       cured    11 bacteriological

hospital_sums <- 
  tb_outcomes %>% 
  group_by(hospital) %>% 
  summarise(cases = sum(cases,na.rm = TRUE),.groups = "drop_last")
hospital_sums

# A tibble: 6 × 2
  hospital         cases
  <chr>            <int>
1 CHPP Akron         875
2 CS Abomey-Calavi   791
3 Hopital Bethesda   256
4 Hopital Savalou     80
5 Hopital St Luc     168
6 St Jean De Dieu    171

ggplot(hospital_sums,aes(x = hospital, y = cases)) +
  geom_col(fill = "steelblue") +
  geom_text(aes(label = cases),vjust = -0.2)

ggplot(hospital_sums,aes(x = hospital, y = cases)) +
  geom_col(fill = "steelblue") +
  geom_text(aes(label = paste(cases, "\ncases")),
            size = 5,
            angle = 0,
            alpha = 0.5,
            color = "black",
            family = "mono",
            fontface = "bold",
            hjust = 0.5,
            vjust = -0.2,
            nudge_y = -10,
            lineheight = 0.8) +
  theme(axis.text.x = element_text(angle = 90))

Labeling Stacked Bar Plots

• We explore labeling in stacked bar plots with two categorical variables.

• We use the tb_outb_outcomes %>% DT::datatable() outcome dataset, summarizing by period_date and diagnosis_type.

tb_outcomes %>% DT::datatable()

tb_sum <- tb_outcomes %>%
  group_by(period_date,diagnosis_type) %>% 
  summarise(cases = sum(cases, na.rm = TRUE),.groups = "drop")
head(tb_sum,5)

# A tibble: 5 × 3
  period_date diagnosis_type  cases
  <chr>       <chr>           <int>
1 2015-01-01  bacteriological   143
2 2015-01-01  clinical           47
3 2015-04-01  bacteriological   163
4 2015-04-01  clinical           35
5 2015-07-01  bacteriological   146

# Summarize the data by period and diagnosis type
# Group by period_date and diagnosis_type
# Summarise cases = Sum(cases)

• Create a simple stacked bar plot using ggplot2

quarter_dx_bar <- 
  ggplot(tb_sum, aes(x = period_date,
                   y = cases,
                   fill = diagnosis_type)) +
  geom_col() +
  labs(title = "New and relapse TB cases per quarter",
       subtitle = "Data from six health facilities in Benin, 2015-2017")
quarter_dx_bar

• We aim to add text labels to each bar segment using cases.

quarter_dx_bar +
  geom_text(aes(label = cases))

• Notice the misalignment of labels - they don’t match bar heights.

• To correct this, set position = “stack” in geom_text().

quarter_dx_bar +
  geom_text(aes(label = cases),
            position = "stack")

• For vertical alignment inside bars, we can use vjust in geom_text()

quarter_dx_bar +
  geom_text(aes(label = cases),
            position = "stack",
            vjust = 1.5)

• But what if we want to center the labels in each bar segment? Can we use position_stack(vjust = 0.5)

quarter_dx_bar +
  geom_text(aes(label = cases),
            position = position_stack(vjust = 0.5))

• This approach is effective for horizontal bar plots - flip axes with coord_flip().

quarter_dx_bar +
  geom_text(aes(label = cases),
            position = position_stack(vjust = 0.5)) +
  coord_flip()

Final product looks like this

quarter_dx_bar +
  geom_text(aes(label = cases),
            position = position_stack(vjust = 0.5),
            color = "white",
            fontface = "bold") +
  coord_flip()

aus_tb_notifs <- 
 aus_tb <- read.csv("C:/Users/Perminus Njiru/OneDrive - LVCT Health/Desktop/Freecodecamp/R/Care_and_Treatment_Analysis/labels_using_ggplot/data/aus_tb_notifs_modified.csv")
aus_tb_notifs %>%
  DT::datatable()

Q: Practice with labeling stacked plots

Create a stacked bar plot showing the distribution per year of TB cases in rural and urban areas using the aus_tb_notifs dataset. Use geom_text() and adjust the position of the labels for clarity.

Hint: Pivot the data so that area_type is a column, then summarize the data by year and area_type, calculating the sum of cases (cases) for each group. The pivoting is done for you in the code below.

# Pivot the data
aus_tb_notifs %>% 
  pivot_longer(cols = c(rural,urban),
               names_to = "area_type",
               values_to = "cases") %>% 
  DT::datatable()

# Pivot the data
aus_tb_notifs %>% 
  pivot_longer(cols = c(rural,urban),
               names_to = "area_type",
               values_to = "cases") %>% 
# Summarize the data by year and area type
group_by(year,area_type) %>% 
  summarise(total_case = sum(cases, na.rm = TRUE),.groups = "drop") %>% 
# Create the stacked bar plot
ggplot(aes(x = year,
           y = total_case,
           fill = area_type)) +
  geom_col() +
  geom_text(
    aes(label = total_case),
    position = position_stack(vjust = 0.5),
    color = "white",
    fontface = "bold",
    size = 3
  ) +
  labs(
    title = "Distribution of TB Cases by Area Type per Year",
    X = "Year",
    y = "Total Cases",
    fill = "Area Type"
  ) + 
  theme_minimal()

Ignoring unknown labels:
• X : "Year"

Labeling Dodged Bar Plots

• Exploring dodged bar charts: multiple categories displayed side by side

• Start by grouping tb_outcomes dataset by hospital and diagnosis_type.

• Calculate the sum of cases (cases) for each group.

hospital_dx_cases <- 
  tb_outcomes %>% 
  group_by(hospital,diagnosis_type) %>% 
  summarise(cases = sum(cases,na.rm = TRUE),.groups = "drop_last")
hospital_dx_cases

# A tibble: 12 × 3
# Groups:   hospital [6]
   hospital         diagnosis_type  cases
   <chr>            <chr>           <int>
 1 CHPP Akron       bacteriological   695
 2 CHPP Akron       clinical          180
 3 CS Abomey-Calavi bacteriological   671
 4 CS Abomey-Calavi clinical          120
 5 Hopital Bethesda bacteriological   139
 6 Hopital Bethesda clinical          117
 7 Hopital Savalou  bacteriological    70
 8 Hopital Savalou  clinical           10
 9 Hopital St Luc   bacteriological   149
10 Hopital St Luc   clinical           19
11 St Jean De Dieu  bacteriological   100
12 St Jean De Dieu  clinical           71

• Create a dodged bar chart, setting position = "dodge" in geom_col()

hospital_dx_bar <- 
  hospital_dx_cases %>% 
  ggplot(aes(x = hospital,
             y = cases,
             fill = diagnosis_type)) +
  geom_col(position = "dodge")

• Annotate chart with geom_text() to display the labels

hospital_dx_bar +
  geom_text(aes(label = cases))

• Not quite right! Need to adjust geom_text() with position = position_dodge() for proper alignment with bars.

hospital_dx_bar +
  geom_text(aes(label = cases),
            position = position_dodge(width = 0.9))

• Finally, shift labels up slightly with vjust for better visualization.

hospital_dx_bar +
  geom_text(aes(label = cases),
            position = position_dodge(width = 0.9),
            vjust = -0.2)

Labeling Percent - stacked Bar Plots

• For percentage stacked bar plot, need to format labels into percentages to match the segments on the chart.

tb_outcomes %>% 
  group_by(hospital, diagnosis_type) %>% 
  summarise(total_cases = sum(cases),.groups = "drop_last") %>% 
  mutate(prop = total_cases / sum(total_cases)) %>% 
  ggplot(aes(x = hospital,
             y = total_cases,
             fill = diagnosis_type)) +
  geom_col(position = "fill") +
  geom_text(aes(label = scales::percent(prop, accuracy = 1)),
            position = position_fill(vjust = 0.5),
            color = "white",
            fontface = "bold",
            size = 4.5)+
  coord_flip()+
  theme_minimal()+
  labs(x = "",
       y = "Proportion",
       fill = "Diagnosis Method",
       title = "New and Relapse Tuberculosis Cases Diagnosis",
       subtitle = "Data from six health facilities in Benin, 2015 - 2017")

• Let’s start with a stacked bar plot showing raw counts.

hosp_dx_sum <- 
  tb_outcomes %>% 
  group_by(hospital,diagnosis_type) %>% 
  summarise(total_cases = sum(cases, na.rm = TRUE),.groups = "drop_last")
hosp_dx_sum

# A tibble: 12 × 3
# Groups:   hospital [6]
   hospital         diagnosis_type  total_cases
   <chr>            <chr>                 <int>
 1 CHPP Akron       bacteriological         695
 2 CHPP Akron       clinical                180
 3 CS Abomey-Calavi bacteriological         671
 4 CS Abomey-Calavi clinical                120
 5 Hopital Bethesda bacteriological         139
 6 Hopital Bethesda clinical                117
 7 Hopital Savalou  bacteriological          70
 8 Hopital Savalou  clinical                 10
 9 Hopital St Luc   bacteriological         149
10 Hopital St Luc   clinical                 19
11 St Jean De Dieu  bacteriological         100
12 St Jean De Dieu  clinical                 71

hosp_dx_sum %>% 
  ggplot(aes(x = hospital,
             y = total_cases,
             fill = diagnosis_type)) +
  geom_col() +
  geom_text(aes(label = total_cases),
            position = "stack")

• Now we can normalize this by using the “fill” position for geom_col() and geom_text(). or rather, position_fill().

hosp_dx_sum %>% 
  ggplot(aes(x = hospital,
             y = total_cases,
             fill = diagnosis_type)) +
  geom_col(position = position_fill()) +
  geom_text(aes(label = total_cases),
            position = position_fill())

• Next, replace raw values with percentages.

hosp_dx_sum <- 
  tb_outcomes %>% 
  group_by(hospital,diagnosis_type) %>% 
  summarise(total_cases = sum(cases, na.rm = TRUE),.groups = "drop_last") %>% 
  group_by(hospital) %>% 
  mutate(prop = total_cases/sum(total_cases))
hosp_dx_sum

# A tibble: 12 × 4
# Groups:   hospital [6]
   hospital         diagnosis_type  total_cases  prop
   <chr>            <chr>                 <int> <dbl>
 1 CHPP Akron       bacteriological         695 0.794
 2 CHPP Akron       clinical                180 0.206
 3 CS Abomey-Calavi bacteriological         671 0.848
 4 CS Abomey-Calavi clinical                120 0.152
 5 Hopital Bethesda bacteriological         139 0.543
 6 Hopital Bethesda clinical                117 0.457
 7 Hopital Savalou  bacteriological          70 0.875
 8 Hopital Savalou  clinical                 10 0.125
 9 Hopital St Luc   bacteriological         149 0.887
10 Hopital St Luc   clinical                 19 0.113
11 St Jean De Dieu  bacteriological         100 0.585
12 St Jean De Dieu  clinical                 71 0.415

• Create a bar chart with the prop column.

hosp_dx_fill <- 
  hosp_dx_sum %>% 
  ggplot(aes(x = hospital,
             y = prop,
             fill = diagnosis_type)) +
  geom_col()

• Add label using geom_text()

hosp_dx_fill +
  geom_text(aes(label = prop),
            position = position_fill())

• Improve label readability by formatting decimals and percentages.

hosp_dx_fill +
  geom_text(aes(label = scales::percent(prop)),
            position = position_fill())

• Center labels using vjust in position_fill().

hosp_dx_fill +
  geom_text(aes(label = scales::percent(prop)),
            position = position_fill(vjust = 0.5))

• Enhance readability with flipped coordinates and aesthetic tweaks.

hosp_dx_fill +
  geom_text(aes(label = scales::percent(prop, accuracy = 1)),
            position = position_fill(vjust = 0.5),
            color = "white",
            fontface = "bold",
            size = 4.5) +
  theme_light() +
  coord_flip()

Labeling Circular Plot

• Begin with summarizing the data by calculating total for each hospital.

total_results <- 
  tb_outcomes %>% 
  group_by(hospital) %>% 
  summarise(total_cases = sum(cases))
total_results

# A tibble: 6 × 2
  hospital         total_cases
  <chr>                  <int>
1 CHPP Akron               875
2 CS Abomey-Calavi         791
3 Hopital Bethesda         256
4 Hopital Savalou           80
5 Hopital St Luc           168
6 St Jean De Dieu          171

• Create a simple bar chart to visualize the data before moving to circular plots.

ggplot(total_results,
       aes(x = 4,
           y = total_cases,
           fill = hospital)) +
  geom_col()

• Transform the bar chart into a basic pie chart using coord_polar().

Outcome_pie <- 
  ggplot(total_results,
       aes(x = 4,
           y = total_cases,
           fill = hospital)) +
  geom_col() +
  theme_void()+
  labs(
    title = "TB Cases in **Rural** and *Urban* Areas",  # bold and italic
    subtitle = "Cases <span style='color:red'>increased</span> in 2020"  # coloured text
  ) +
  theme(
    plot.title = element_markdown(),    # enables markdown in title
    plot.subtitle = element_markdown()  # enables markdown in subtitle
  )+
  coord_polar(theta = "y")
Outcome_pie

• Next, focus on adding labels to the pie chart.

Outcome_pie +
  geom_text(aes(label = total_cases),
            position = position_stack(vjust = 0.5))

• Create a base donut chart with xlim().

Outcome_donut <- 
  ggplot(total_results,
       aes(x = 4,
           y = total_cases,
           fill = hospital)) +
  geom_col() +
  theme_void()+
  labs(
    title = "TB Cases in **Rural** and *Urban* Areas",  # bold and italic
    subtitle = "Cases <span style='color:red'>increased</span> in 2020"  # coloured text
  ) +
  theme(
    plot.title = element_markdown(),    # enables markdown in title
    plot.subtitle = element_markdown()  # enables markdown in subtitle
  )+
  xlim(c(0,4.5))+
  coord_polar(theta = "y")
Outcome_donut

• Next, focus on adding labels to the donut chart.

Outcome_donut +
  geom_text(aes(label = total_cases),
            position = position_stack(vjust = 0.5))

• Enhance the chart’s aesthetics with theme_void(), color adjustments, and label styling.

Outcome_donut +
  geom_text(aes(label = total_cases),
            position = position_stack(vjust = 0.5),
            color = "white",
            fontface = "bold") +
  scale_fill_viridis_d()

# scale_fill_viridis_d(option = "viridis")  # default - purple to yellow
Outcome_pie +
  geom_label(aes(label = total_cases),
            position = position_stack(vjust = 0.5),
            color = "white",
            fontface = "bold",
            show.legend = FALSE) +
  scale_fill_viridis_d()

# scale_fill_viridis_d(option = "magma")    # black to yellow
Outcome_pie +
  geom_label(aes(label = total_cases),
            position = position_stack(vjust = 0.5),
            color = "white",
            fontface = "bold",
            show.legend = FALSE) +
  scale_fill_viridis_d(option = "magma")

# scale_fill_viridis_d(option = "plasma")   # purple to yellow-pink
Outcome_pie +
  geom_label(aes(label = total_cases),
            position = position_stack(vjust = 0.5),
            color = "white",
            fontface = "bold",
            show.legend = FALSE) +
  scale_fill_viridis_d(option = "plasma")

# scale_fill_viridis_d(option = "inferno")  # black to yellow-white
Outcome_pie +
  geom_label(aes(label = total_cases),
            position = position_stack(vjust = 0.5),
            color = "white",
            fontface = "bold",
            show.legend = FALSE) +
  scale_fill_viridis_d(option = "inferno")

# scale_fill_viridis_d(option = "cividis")  # blue to yellow
Outcome_pie +
  geom_label(aes(label = total_cases),
            position = position_stack(vjust = 0.5),
            color = "white",
            fontface = "bold",
            show.legend = FALSE) +
  scale_fill_viridis_d(option = "cividis")

Ask yourself: what type is my fill variable?

1. Words/names? → scale_fill_viridis_d()

2. Numbers (smooth)? → scale_fill_viridis_c()

3. Numbers (grouped)? → scale_fill_viridis_b()

Consider the example below, which uses for the plot title, subtitle and bar labels:

# Data and plot

medicaldata::strep_tb %>% 
  count(gender) %>% 
  mutate(gender_label = paste0(
  "<b><span style='font-size:16pt'>", n, "</span></b> ",
  if_else(gender == "M", "men", "women"))) %>% 
  ggplot(aes(x = gender,
             fill = gender,
             y = n)) +
  geom_col() +
  scale_fill_manual(values = c("M" = "#ee6c4d", "F" = "#424874")) +
  labs(
    title = "<b><span style='color:#424874; font-size:19pt'>Female</span> vs 
    <span style='color:#ee6c4d; font-size:19pt'>Male</span> 
    Patients in Strep Study</b>",
    subtitle = "<span style='color:gray60'>A demonstration of custom text labels with 
</span>**{ggtext}**") +
  theme_classic() +
  theme(plot.title = element_textbox_simple(),
        plot.subtitle = element_textbox_simple(),
        legend.position = "none",
        axis.text.x = element_blank()) +
  geom_richtext(aes(label = gender_label, y = n/2),
                label.r = grid::unit(5, "pt"), fill = "white")