How to only label the outliers in a ggplot graph with R

Another blog I will make to have easy access to code snippets for my own record.

We will use an example with data from V-DEM.

Click here to read more about downloading the V-DEM dataset

How to download and animate the Varieties of Democracy (V-DEM) dataset in R

string_vector <- c("_nr", "_codehigh", "_codelow", "_mean", "_sd", "_ord",
                   "_ord_codehigh", "_ord_codelow", "_osp", "_osp_codehigh",
                   "_osp_codelow", "_osp_sd")

pattern <- paste(string_vector, collapse = "|")

vdem %<>% 
  dplyr::select(-matches(pattern))


vdem %<>% 
  filter(year %in% c(1900:2022))

v2x_jucon: To what extent does the executive respect the constitution and comply with court rulings, and to what extent is the judiciary able to act in an independent fashion?

v2x_corr: : How pervasive is political corruption?

vdem %>%
  group_by(country_name) %>% 
  summarise(avg_corr = mean(v2x_corr, na.rm = TRUE),
            avg_judic = mean(v2x_jucon, na.rm = TRUE)) -> vdem_summarised
  ggplot(aes(x = avg_corr, 
             y = avg_judic)) + 
  geom_point(alpha = 0.6) +
  scale_y_continuous(labels = scales::label_comma()) 

First we need to model the variables in a linear regression

Then we create a residuals variable

And then we can find what residuals are two standard deviations from the OLS line

  model <- lm(avg_judic ~ avg_corr, data = vdem_summarised)
  
  vdem_summarised <- vdem_summarised %>%
    mutate(residuals = resid(model))
  
  residual_threshold <- 2 * sd(vdem_summarised$residuals)

Next we flag certain countries as outliers based on the model residuals

  vdem_summarised <- vdem_summarised %>%
    mutate(outlier = ifelse(abs(residuals) > residual_threshold, TRUE, FALSE))

And we plot it out

 vdem_summarised %>%
    ggplot(aes(x = avg_corr, y = avg_judic)) +
    geom_smooth(color = "#003d5b", 
                method = "lm", 
                se = FALSE,
                size = 3,
                alpha = 0.2) +
    geom_point(aes(color = outlier), 
               size = 4, 
               alpha = 0.6,
               ) + 
    scale_color_manual(values = c("FALSE" = "#00798c", 
                                  "TRUE" = "#c1121f")) +
    ggrepel::geom_label_repel(
      data = filter(vdem_summarised, outlier),
      aes(label = country_name),
      size = 3, 
      nudge_x = 0.1, 
      nudge_y = 0.1, 
      color = "#c1121f"
    ) +
    theme_minimal() + 
    labs(title = "Judicial Independence vs. Political Corruption",
         caption = "V-DEM average 1900 - 2020",
         x = "Average Judicial Independence",
         y = "Average Political Corruption") +
    guides(color = guide_legend(override.aes = list(size = 4))) + 
    theme(text = element_text(size = 12),  # Default text size for all text elements
          plot.title = element_text(size = 20, face="bold"),  # Plot title
          axis.title = element_text(size = 16),  # Axis titles (both x and y)
          axis.text = element_text(size = 14),  # Axis text (both x and y)
          legend.title = element_text(size = 14),  # Legend title
          legend.text = element_text(size = 12))  # Legend items
 

First, we will calculate the inter-quartile range and define outliers for our y variable.

iqr <- IQR(df$y)
upper_threshold <- quantile(df$y, 0.75) + 1.5 * iqr
lower_threshold <- quantile(df$y, 0.25) - 1.5 * iqr

And we use that iqr data.frame and add it to the df

df <- df %>%
  mutate(outlier = ifelse(y < lower_threshold | y > upper_threshold, TRUE, FALSE))

And we can graph the ggplot with the geom_text() only of the outliers.

ggplot(df, aes(x = x, y = y)) +
  geom_point() +
  ggrepel::geom_label_repel(data = filter(df, outlier), aes(label = country_name), nudge_y = 0.25) +
  theme_minimal()