KEY - visualizing distributions

Author

Your Name

Published

January 14, 2026

Note

This key follows the visualizing distributions slides. Please be sure to cross-reference the slides, which contain important information and additional context!

Setup

##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
##                                    setup                                 ----
##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

#..........................load packages.........................
library(tidyverse)
library(chron)
library(naniar)
library(ggridges)
library(gghighlight)
library(ggbeeswarm)
library(palmerpenguins) # for some minimal examples

#..........................import data...........................
mko <- read_csv(here::here("week2", "data", "mohawk_mooring_mko_20250117.csv"))

##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
##                                wrangle data                              ----
##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

mko_clean <- mko |>

  # keep only necessary columns ----
  select(year, month, day, decimal_time, Temp_bot, Temp_top, Temp_mid) |>

  # create datetime column (not totally necessary for our plots, but it can be helpful to know how to do this!) ----
  unite(date, year, month, day, sep = "-", remove = FALSE) |>
  mutate(time = chron::times(decimal_time)) |>
  unite(date_time, date, time, sep = " ") |>

  # coerce data types ----
  mutate(date_time = as_datetime(date_time, "%Y-%m-%d %H:%M:%S", tz = "GMT"), 
         year = as.factor(year),
         month = as.factor(month),
         day = as.numeric(day)) |>

  # add month name by indexing the built-in `month.name` vector ----
  mutate(month_name = month.name[month]) |> 

  # replace 9999s with NAs ----
  naniar::replace_with_na(replace = list(Temp_bot = 9999, 
                                         Temp_top = 9999, 
                                         Temp_mid = 9999)) |>

  # select/reorder desired columns ----
  select(date_time, year, month, day, month_name, Temp_bot, Temp_mid, Temp_top)

##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
##                            explore missing data                          ----
##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

# Missing data can have unexpected effects on your analyses
# It's important to explore your data for missing values so that you can decide if and how to handle them
# Data loggers can be prone to missingness (e.g. full memory, dead batteries, replacement logger)
# We can use {naniar} to explore the frequency and patterns in missing data
# Below is a short example of some naniar tools for doing so
# Check out The Missing Book (https://tmb.njtierney.com/) by Nick Tierney and Allison Horst for some great guidance

#..........counts & percentages of missing data by year..........
see_NAs <- mko_clean |> 
  group_by(year) |> 
  naniar::miss_var_summary() |>
  filter(variable == "Temp_bot")

#...................visualize missing Temp_bot...................
bottom <- mko_clean |> select(Temp_bot)
missing_temps <- naniar::vis_miss(bottom)

Histograms & Density Plots

  • great for visualizing the distribution of a numeric variable with one to a few groups

Distinction

  • Histograms show the counts (frequency) of values in each range (bin), represented by the height of the bars.
  • Density plots show the relative proportion of values across the range of a variable. The total area under the curve equals 1, and peaks indicate values are more concentrated. Density plots do not show the absolute number of observations.
#.....................create some dummy data.....................
dummy_data <- data.frame(value = c(rnorm(n = 100, mean = 5),
                                   rnorm(n = 200, mean = 10)),
                         group = rep(c("A", "B"),
                                     times = c(100, 200)))

#............................histogram...........................
ggplot(dummy_data, aes(x = value, fill = group)) +
  geom_histogram(position = "identity", alpha = 0.7) +
  geom_rug(aes(color = group), alpha = 0.75)
#..........................density plot..........................
ggplot(dummy_data, aes(x = value, fill = group)) +
  geom_density(alpha = 0.7) +
  geom_rug(aes(color = group), alpha = 0.75)

Avoid plotting too many groups

#..................histogram with all 12 months..................
mko_clean |> 
  mutate(month_name = factor(month_name, levels = month.name)) |> 
  ggplot(aes(x = Temp_bot, fill = month_name)) +
  geom_histogram(position = "identity", alpha = 0.5)
#................density plot with all 12 months.................
mko_clean |> 
  mutate(month_name = factor(x = month_name, levels = month.name)) |> 
  ggplot(aes(x = Temp_bot, fill = month_name)) +
  geom_density(alpha = 0.5)

Consider faceting (small multiples)

#...................histogram faceted by month...................
mko_clean |> 
  mutate(month_name = factor(month_name, levels = month.name)) |> 
  ggplot(aes(x = Temp_bot)) +
  geom_histogram() +
  facet_wrap(~month_name)
#..................density plot faceted by month.................
mko_clean |> 
  mutate(month_name = factor(month_name, levels = month.name)) |> 
  ggplot(aes(x = Temp_bot)) +
  geom_density(fill = "gray30") +
  facet_wrap(~month_name)

Consider plotting fewer groups + update colors

  • histograms & density plots are best suited for visualizing the distribution of one to a small number of groups
#..................histogram with fewer months...................
mko_clean |> 
  mutate(month_name = factor(month_name, levels = month.name)) |> 
  filter(month_name %in% c("April", "June", "October")) |> 
  ggplot(aes(x = Temp_bot, fill = month_name)) + 
  geom_histogram(position = "identity", alpha = 0.5, color = "black", binwidth = 1) +
  scale_fill_manual(values = c("#2C5374", "#ADD8E6", "#8B3A3A"))
#.................density plot with fewer months.................
mko_clean |> 
  filter(month_name %in% c("April", "June", "October")) |> 
  ggplot(aes(x = Temp_bot, fill = month_name)) +
  geom_density(alpha = 0.5) + 
  scale_fill_manual(values = c("#2C5374", "#ADD8E6", "#8B3A3A"))

Question:

Why are the months in our density plot still in chronological order, despite not reordering them using mutate() (as we do for our histogram)?

Modify bin / band widths

  • consider your scale of interest e.g. does a bin width of 1 (°C) make sense for our histogram? Does your density plot’s bandwidth show the true underlying data distribution shape while reducing noise?
#...............histogram with modified bin widths...............
mko_clean |> 
  filter(month_name %in% c("April", "June", "October")) |> 
  mutate(month_name = factor(month_name, levels = month.name)) |> 
  ggplot(aes(x = Temp_bot, fill = month_name)) +
  geom_histogram(position = "identity", alpha = 0.5, binwidth = 1) +
  scale_fill_manual(values = c("#2C5374", "#ADD8E6", "#8B3A3A"))
#..............density plot with modified bin widths.............
mko_clean |> 
  filter(month_name %in% c("April", "June", "October")) |> 
  ggplot(aes(x = Temp_bot, fill = month_name)) +
  geom_density(alpha = 0.5, adjust = 0.5) + 
  scale_fill_manual(values = c("#2C5374", "#ADD8E6", "#8B3A3A"))

Overlay histogram & density plots as a sanity check

Question:

What should you carefully consider when checking the smoothing asumptions of your density curve against a histogram?

#..................overlay histogram & density...................
ggplot(mko_clean, aes(x = Temp_bot, y = after_stat(density))) + # scale down hist to match density curve
  geom_histogram(fill = "gray", color = "black", alpha = 0.75) +
  geom_density(linewidth = 1)

Ridgeline plots

  • great for comparing distribution shapes of a numeric variable across many ordered groups
#......................basic ridgeline plot......................
mko_clean |> 
  mutate(month_name = factor(month_name, levels = rev(month.name))) |> # alt, within ggplot: `scale_y_discrete(limits = rev(month.name))`
  ggplot(aes(x = Temp_bot, y = month_name)) +
  ggridges::geom_density_ridges(rel_min_height = 0.01, scale = 3)  # `rel_min_height` sets threshold for relative height of density curves (any values below threshold treated as 0); `scale` controls extent to which different densities overlap
#....................fill with color gradient....................
mko_clean |> 
  mutate(month_name = factor(month_name, levels = rev(month.name))) |> 
  ggplot(aes(x = Temp_bot, y = month_name, fill = after_stat(x))) + # `fill = after_stat(x)` tells ggplot to compute the `x` values (representing `Temp_bot`) after the statistical transformation (density estimation) and map those computed `x` values to the `fill` aesthetic. As a result, the gradient fill of each density curve will reflect the temperature values along the x-axis.
  ggridges::geom_density_ridges_gradient(rel_min_height = 0.01, scale = 3) +
  scale_fill_gradientn(colors = c("#2C5374","#849BB4", "#D9E7EC", "#EF8080", "#8B3A3A"))

Boxplots

  • great for summarizing the distribution of a numeric variable for multiple to many groups
#...................boxplot with all 12 months...................
ggplot(mko_clean, aes(x = month_name, y = Temp_bot)) +
  geom_boxplot() +
  scale_x_discrete(limits = rev(month.name)) + # an alt way to reorder groups within ggplot, rather than during data wrangling
  coord_flip()

Question:

What are the tradeoffs between reordering groups within a ggplot (as above) vs. during the data wrangling stage (e.g. as we did for our histogram and density and ridgeline plots)?

Highlight group(s) of interest to focus attention

#..................highlight a particular group..................
ggplot(mko_clean, aes(x = month_name, y = Temp_bot, fill = month_name)) +
  geom_boxplot() +
  scale_x_discrete(limits = rev(month.name)) +
  gghighlight::gghighlight(month_name == "October") +
  coord_flip() +
  theme(legend.position = "none")

Overlay jittered data, if appropriate

#........does not work if you have too many observations.........
ggplot(mko_clean, aes(x = month_name, y = Temp_bot)) +
  geom_boxplot(outlier.shape = NA) + # remember to remove outliers!
  geom_jitter(alpha = 0.5, width = 0.2) +
  scale_x_discrete(limits = rev(month.name)) + # an alt way to reorder groups within ggplot, rather than during data wrangling
  coord_flip()

#...........but great when you have fewer observations...........
ggplot(penguins, aes(x = species, y = body_mass_g)) +
  geom_boxplot(outlier.shape = NA) + # remember to remove outliers!
  geom_jitter(alpha = 0.5, width = 0.2)

Beeswarm plots as an alternative

#.......................boxplot + beeswarm.......................
ggplot(penguins, aes(x = species, y = body_mass_g)) +
  geom_boxplot(outlier.shape = NA) + # remember to remove outliers!
  ggbeeswarm::geom_beeswarm(size = 1)

Dodge when you have an additional grouping variable

#......................dodge groups by year......................
ggplot(penguins, aes(x = species, y = body_mass_g, color = as.factor(year))) +
  geom_boxplot(outlier.shape = NA) + # remember to remove outliers!
  geom_point(alpha = 0.5, 
             position = position_jitterdodge(jitter.width = 0.2))

Violin plots

  • great for visualizing the shape of the distribution of a numeric variable for multiple to many groups
  • consider overlaying another chart type for added context
#........................violin + box plot.......................
ggplot(mko_clean, aes(x = month_name, y = Temp_bot)) +
  geom_violin() +
  geom_boxplot(width = 0.1, fill = "gray", color = "gray30", alpha = 0.5) +
  scale_x_discrete(limits = rev(month.name)) +
  coord_flip()