How To Plot Histogram In R

A histogram remains one of the most fundamental tools for visualizing the distribution of a single continuous variable. In R, creating this visualization can range from a single line of base code to a highly customized graphic built with the ggplot2 package. Understanding the nuances of bin widths, density scaling, and aesthetic mapping allows analysts to move beyond default outputs and produce publication-ready figures that reveal the true shape of their data The details matter here..

Getting Started with Base R Graphics

The simplest way to generate a histogram in R requires no external packages. Because of that, the hist() function is built into the graphics package, which loads automatically when you start an R session. This makes it ideal for quick exploratory data analysis (EDA) where speed is the priority Worth knowing..

People argue about this. Here's where I land on it.

Consider the built-in mtcars dataset. To visualize the distribution of miles per gallon (mpg), the syntax is remarkably concise:

hist(mtcars$mpg)

Executing this code opens a graphics device displaying a frequency histogram. Which means by default, R calculates bin widths using the Sturges algorithm, labels the y-axis as "Frequency," and titles the plot with the variable name. While functional, the default output often lacks the polish required for reports Nothing fancy..

Worth pausing on this one That's the part that actually makes a difference..

You can immediately improve readability by adding arguments directly into the function call. That said, the main, xlab, and ylab arguments handle titles and axis labels. The col and border arguments control the fill color and the outline of the bars, respectively Took long enough..

hist(mtcars$mpg,
     main = "Distribution of Miles Per Gallon",
     xlab = "Miles Per Gallon (MPG)",
     ylab = "Frequency",
     col = "steelblue",
     border = "white")

This enhanced version produces a cleaner, self-explanatory chart suitable for internal presentations or rapid sharing with colleagues.

Controlling Bins and Breakpoints

The most critical decision when constructing a histogram is determining the number and width of bins. Too few bins oversimplify the distribution, hiding multimodality or skewness. Too many bins introduce noise, obscuring the underlying pattern. Base R offers the breaks argument to control this behavior precisely.

Counterintuitive, but true.

You can pass a single integer to breaks as a suggestion for the number of bins, though the algorithm may adjust this to create "pretty" round numbers for bin edges. For exact control, provide a vector of specific breakpoints That's the part that actually makes a difference..

# Suggest 10 bins
hist(mtcars$mpg, breaks = 10, col = "lightgray")

# Define exact breakpoints (e.g., 10, 15, 20, 25, 35)
hist(mtcars$mpg, breaks = c(10, 15, 20, 25, 30, 35), col = "lightgray")

Alternatively, you can specify a binning algorithm by name. That's why common options include "Sturges" (the default), "Scott", and "FD" (Freedman-Diaconis). The Freedman-Diaconis rule is often preferred for larger datasets or data with outliers because it bases bin width on the interquartile range (IQR), making it more solid than Sturges, which relies only on sample size.

hist(mtcars$mpg, breaks = "FD", col = "lightcoral")

Experimenting with different binning strategies is a standard part of the EDA workflow. It ensures that the visual story you tell is not an artifact of an arbitrary default setting.

Plotting Density Instead of Frequency

In many statistical contexts, comparing a histogram to a theoretical probability density function (like the Normal curve) is necessary. Now, a frequency histogram uses counts on the y-axis, making the total area dependent on the sample size. A density histogram scales the y-axis so that the total area of all bars sums to 1 Worth keeping that in mind..

In base R, switch to density mode using freq = FALSE (or probability = TRUE).

hist(mtcars$mpg,
     freq = FALSE,
     main = "Density Histogram of MPG",
     xlab = "MPG",
     col = "lightgreen",
     border = "white")

# Overlay a kernel density estimate
lines(density(mtcars$mpg), col = "red", lwd = 2)

# Overlay theoretical normal curve
xfit <- seq(min(mtcars$mpg), max(mtcars$mpg), length = 40)
yfit <- dnorm(xfit, mean = mean(mtcars$mpg), sd = sd(mtcars$mpg))
lines(xfit, yfit, col = "blue", lwd = 2, lty = 2)

This code produces a plot where the red line represents the empirical kernel density estimate (KDE) and the blue dashed line represents the theoretical Normal distribution. This visual comparison is a powerful method for assessing normality assumptions required by many parametric tests Less friction, more output..

Creating Professional Histograms with ggplot2

While base graphics are fast, the ggplot2 package (part of the tidyverse) is the industry standard for creating complex, layered, and aesthetically refined visualizations. It implements the Grammar of Graphics, separating data, aesthetics, and geometry.

First, ensure the package is installed and loaded:

install.packages("ggplot2")
library(ggplot2)

The fundamental syntax maps variables to aesthetics (aes) and adds geoms (geometric objects). For a histogram, the geom is geom_histogram().

ggplot(data = mtcars, aes(x = mpg)) +
  geom_histogram()

By default, ggplot2 uses 30 bins and a dark grey fill. It also prints a message suggesting you pick a better binwidth. This verbosity is a feature, not a bug—it forces the analyst to make a conscious decision about binning Practical, not theoretical..

Customizing Bins and Appearance in ggplot2

You control binning via binwidth (exact width), bins (total number), or breaks (exact boundaries). Aesthetics like fill, color (border), and alpha (transparency) are mapped outside aes() if they are constant, or inside aes() if they vary by group.

ggplot(mtcars, aes(x = mpg)) +
  geom_histogram(binwidth = 2, fill = "steelblue", color = "white", alpha = 0.8) +
  labs(title = "MPG Distribution (ggplot2)",
       subtitle = "Binwidth set to 2 MPG",
       x = "Miles Per Gallon",
       y = "Count") +
  theme_minimal()

The theme_minimal() call strips away the grey background and gridlines typical of the default theme, producing a cleaner look often preferred in modern data science reports. Other popular themes include theme_classic(), theme_bw(), and theme_light().

Mapping Aesthetics to Variables

One of ggplot2's superpowers is mapping visual properties to data columns. To give you an idea, you can fill bars based on a categorical variable like transmission type (am: 0 = automatic, 1 = manual) to create a stacked or dodged histogram.

# Convert 'am' to a factor with meaningful labels for the legend
mtcars$am <- factor(mtcars$am, labels = c("Automatic", "Manual"))

# Stacked histogram (default position)
ggplot(mtcars, aes(x = mpg, fill = am)) +
  geom_histogram(binwidth = 2, color = "white", position = "stack") +
  scale_fill_brewer(palette = "Set1") +
  labs(fill = "Transmission")

# Dodged (side-by-side

To create a side-by-side comparison, you simply change the `position` argument from `"stack"` to `"dodge"`. This is often preferable when comparing the overall shape and spread of the distributions, as stacking can obscure the details of smaller groups.

```r
# Dodged (side-by-side) histogram
ggplot(mtcars, aes(x = mpg, fill = am

The missing line can be completed as follows, and the `position` argument switched to `"dodge"` to obtain a side‑by‑side view:

```r
# Dodged (side‑by‑side) histogram
ggplot(mtcars, aes(x = mpg, fill = am)) +
  geom_histogram(binwidth = 2, colour = "white", position = "dodge") +
  scale_fill_brewer(palette = "Set1") +
  labs(fill = "Transmission",
       title = "MPG Distribution by Transmission Type",
       x = "Miles Per Gallon",
       y = "Count") +
  theme_minimal()

When position = "dodge" is used, each level of the grouping variable (am in this case) receives its own set of bars that are placed next to each other rather than stacked. This makes it easier to compare the shapes of the two distributions directly. If the counts become difficult to read because the bars are too narrow, you can increase the binwidth or let ggplot2 choose the optimal width by omitting the argument entirely Worth keeping that in mind..

Fine‑tuning the Appearance

• Colors – Instead of relying on a built‑in palette, you can supply a custom vector of colors with scale_fill_manual():

  scale_fill_manual(values = c("Automatic" = "#1b9e77", "Manual" = "#d95f02"))
  

• Transparency – Adding a modest alpha value (e.g., 0.6) can help overlapping bars remain visible, especially when the groups have similar ranges Worth keeping that in mind..

• Axis limits – To prevent the y‑axis from stretching unnecessarily, use coord_cartesian(ylim = c(0, 8)) or simply set ylim inside labs() with fill = "Count".

• Faceting – When you have more than one categorical dimension, facet_wrap() or facet_grid() lets you create separate panels for each combination. For a histogram, this often means faceting by a factor such as cyl (number of cylinders):

  ggplot(mtcars, aes(x = mpg, fill = am)) +
    geom_histogram(binwidth = 2, colour = "white", position = "dodge") +
    facet_wrap(~ cyl) +
    scale_fill_brewer(palette = "Set1") +
    labs(title = "MPG by Cylinders and Transmission",
         x = "Miles Per Gallon",
         y = "Count",
         fill = "Transmission") +
    theme_minimal()
  

From Histogram to Density

If the interest lies in the shape of the distribution rather than the raw counts, replace the default y aesthetic with after_stat(density). This transformation rescales the histogram so that the area under the curve equals one, enabling direct comparison with smooth density estimates Worth keeping that in mind..

ggplot(mtcars, aes(x = mpg, fill = am)) +
  geom_histogram(binwidth = 2, colour = "white", position = "dodge",
                 aes(y = after_stat(density))) +
  scale_fill_brewer(palette = "Set1") +
  labs(title = "Density of MPG by Transmission",
       x = "Miles Per Gallon",
       y = "Density",
       fill = "Transmission") +
  theme_minimal()

The resulting plot shows two density curves that can be overlaid with geom_density() for a smoother visual representation The details matter here. Turns out it matters..

Saving and Exporting

Once the visual meets your standards, you can export it in a variety of formats. The ggsave() function automatically detects the file type from the extension and respects the current plot dimensions.

ggsave("mpg_histogram.png", width = 8, height = 5, dpi = 300)

For vector graphics, use .Think about it: pdf or . svg, which preserve sharpness at any size and are ideal for publications That alone is useful..

Concluding Remarks

ggplot2 transforms a simple histogram into a fully customizable visual story by separating data, aesthetics, and geometry. Which means through the aes() mapping, the geom_histogram() layer, and a rich set of options for binning, colour, positioning, and faceting, analysts can craft graphics that are both informative and aesthetically refined. Whether the goal is to juxtapose transmission types, explore density, or embed the result in a report, the combination of ggplot2’s layered grammar and the flexibility of the underlying data frame empowers users to communicate insights with clarity and style.

Counterintuitive, but true Simple, but easy to overlook..