Introduction

ggplot2 implements the Grammar of Graphics, a systematic approach to building plots layer-by-layer. Every plot follows a consistent structure:

  1. Data – the dataset
  2. Aesthetics (aes) – map variables to visual properties (x, y, colour, size)
  3. Geoms – geometric objects (geom_point(), geom_bar(), …)
  4. Scales – control mapping from data to visual space
  5. Facets – sub-plots by categorical variables

This document walks through seven essential chart types, explains when to use each, and shows how to implement them.

1. Scatter Plots

Utility

  • Purpose: Visualise the relationship between two continuous variables.
  • What to look for: Direction (positive/negative), form (linear/non-linear), strength (tight vs. loose cluster), and outliers.
  • Best for: Hypothesis generation about correlation, trend detection, and identifying data gaps.

Implementation

geom_point() is the workhorse. Additional layers like geom_smooth() add trend lines.

ggplot(mtcars, aes(x = wt, y = mpg)) +
  geom_point(aes(color = factor(cyl)), size = 3, alpha = 0.8) +
  geom_smooth(method = "lm", se = TRUE, color = "black", linetype = "dashed") +
  scale_color_viridis_d(name = "Cylinders") +
  labs(
    title = "Vehicle Weight vs. Fuel Efficiency",
    subtitle = "Coloured by cylinder count with linear trend line",
    x = "Weight (1000 lbs)",
    y = "Miles per Gallon"
  )

Key arguments: alpha for transparency (handles over-plotting), size for point emphasis, and color/fill for grouping.

2. Histograms

Utility

  • Purpose: Show the distribution of a single continuous variable by binning values.
  • What to look for: Skewness (tail direction), modality (number of peaks), spread, and gaps.
  • Best for: Initial data exploration, detecting data entry errors, and understanding central tendency.

Implementation

geom_histogram() counts observations per bin. Choose binwidth carefully – too wide hides detail, too narrow creates noise.

ggplot(mtcars, aes(x = mpg)) +
  geom_histogram(binwidth = 2, fill = "steelblue", color = "white", alpha = 0.9) +
  labs(
    title = "Distribution of Miles per Gallon",
    subtitle = "Binwidth = 2",
    x = "MPG",
    y = "Count"
  )

Tip: Overlay a density curve (next section) by setting aes(y = after_stat(density)) to compare the histogram with a smooth estimate.

3. Density Plots

Utility

  • Purpose: A smoothed version of the histogram – estimates the probability density function.
  • What to look for: Same as histograms, but easier to compare multiple groups on the same axes without bin-artefacts.
  • Best for: Comparing distributions across categories (e.g., MPG by number of cylinders).

Implementation

geom_density() uses a kernel smoother. The adjust parameter controls smoothness (higher = smoother).

ggplot(mtcars, aes(x = mpg, fill = factor(cyl))) +
  geom_density(alpha = 0.5, adjust = 1.5) +
  scale_fill_viridis_d(name = "Cylinders") +
  labs(
    title = "Density of MPG by Cylinder Count",
    subtitle = "Transparent overlap highlights differences in distribution shape",
    x = "MPG",
    y = "Density"
  )

When to prefer over histograms: When sample size is small, or you need a clean visual for overlapping groups.

4. Boxplots

Utility

  • Purpose: Display the five-number summary (min, Q1, median, Q3, max) and outliers.
  • What to look for: Median differences, interquartile range (IQR), symmetry of the middle 50%, and extreme points.
  • Best for: Comparing distributions across several categorical groups in a compact, statistically robust way.

Implementation

geom_boxplot() – whiskers extend to 1.5×IQR; beyond that are plotted as points.

ggplot(mtcars, aes(x = factor(cyl), y = mpg, fill = factor(cyl))) +
  geom_boxplot(outlier.color = "red", outlier.size = 2.5, alpha = 0.8) +
  scale_fill_viridis_d(name = "Cylinders") +
  labs(
    title = "MPG Distribution by Cylinder Count",
    subtitle = "Red points indicate potential outliers",
    x = "Cylinders",
    y = "MPG"
  ) +
  theme(legend.position = "none")

Utility in practice: Quickly see if groups have different medians, spreads, and whether outliers exist.

5. Violin Plots

Utility

  • Purpose: Combine a boxplot with a mirrored density plot – shows the full shape of the distribution and summary statistics.
  • What to look for: Multi-modality (multiple peaks) within groups, which boxplots hide.
  • Best for: Rich group comparisons when you suspect complex distributions (e.g., bimodal data).

Implementation

geom_violin() with trim = FALSE keeps the tails. Add draw_quantiles to overlay quartile lines.

ggplot(mtcars, aes(x = factor(cyl), y = mpg, fill = factor(cyl))) +
  geom_violin(trim = FALSE, draw_quantiles = c(0.25, 0.5, 0.75), alpha = 0.7) +
  scale_fill_viridis_d(name = "Cylinders") +
  labs(
    title = "Violin Plot of MPG by Cylinders",
    subtitle = "Quartile lines (25%, 50%, 75%) are superimposed",
    x = "Cylinders",
    y = "MPG"
  ) +
  theme(legend.position = "none")

Combination tip: Overlay geom_boxplot(width = 0.1) on top of a violin to keep the summary visible while showing the density.

6. Heatmaps

Utility

  • Purpose: Encode a matrix of numeric values using colour intensity.
  • What to look for: Patterns, clusters, and gradients across two categorical/discretised dimensions.
  • Best for: Visualising aggregated tables (e.g., average sales by region and quarter), confusion matrices, or any 2-D grid of values.

Implementation

geom_tile() creates rectangular cells. Use geom_text() to overlay actual numbers if the grid is small.

# Create aggregated data: average MPG by gear and cylinder
heat_data <- mtcars %>%
  group_by(gear, cyl) %>%
  summarise(avg_mpg = mean(mpg), .groups = "drop")

ggplot(heat_data, aes(x = factor(gear), y = factor(cyl), fill = avg_mpg)) +
  geom_tile(color = "white", linewidth = 1) +
  geom_text(aes(label = round(avg_mpg, 1)), color = "black", size = 5) +
  scale_fill_viridis_c(option = "plasma", name = "Avg MPG") +
  labs(
    title = "Average MPG by Gear and Cylinders",
    subtitle = "Darker colours indicate higher fuel efficiency",
    x = "Number of Gears",
    y = "Cylinders"
  )

Note: Use coord_fixed() if you need square tiles, especially when both axes have the same scale.

7. Correlation Matrices

Utility

  • Purpose: Visualise pairwise correlations between multiple numeric variables.
  • What to look for: Strong positive (red) / negative (blue) relationships at a glance.
  • Best for: Multicollinearity diagnosis in regression modelling, and initial feature selection.

Implementation

Compute cor() on numeric columns, melt() into long format, then plot with a diverging colour scale (midpoint = 0).

corr_matrix <- round(cor(mtcars), 2)
corr_melt <- melt(corr_matrix, varnames = c("Var1", "Var2"))

ggplot(corr_melt, aes(x = Var1, y = Var2, fill = value)) +
  geom_tile(color = "white") +
  geom_text(aes(label = value), size = 3.5, color = "black") +
  scale_fill_gradient2(
    low = "#2c3e50",    # dark blue for negative
    mid = "white",      # zero
    high = "#e74c3c",   # red for positive
    midpoint = 0,
    limit = c(-1, 1),
    name = "Correlation"
  ) +
  labs(
    title = "Correlation Matrix of mtcars Variables",
    subtitle = "Values range from -1 (negative) to +1 (positive)"
  ) +
  theme(
    axis.text.x = element_text(angle = 45, hjust = 1),
    axis.title = element_blank()
  )

Why diverging colours? They make the sign and magnitude immediately obvious. scale_fill_gradient2 is perfect for this.

Quick Comparison Table

Plot Geom Data Types Primary Utility
Scatter geom_point() 2 continuous Relationship, trends, clustering, outliers
Histogram geom_histogram() 1 continuous Distribution shape, skewness, modality
Density geom_density() 1 continuous (+ fill groups) Smoothed distribution, compare multiple groups
Boxplot geom_boxplot() 1 continuous + 1 categorical Summarise and compare distributions (medians, IQR)
Violin geom_violin() 1 continuous + 1 categorical Full distribution + summary for group comparisons
Heatmap geom_tile() 2 categorical (values) Visualise 2-D aggregated metrics / patterns
Correlation geom_tile() Many numeric → matrix Spot collinearity, feature relationships at scale

Conclusion

Mastering these seven visualisations covers 90% of day-to-day exploratory data analysis needs. Remember the layered philosophy of ggplot2:

Data → Aesthetics → Geoms → Scales → Facets → Themes

Start simple, then iteratively add labels, colors, and statistical layers to tell a compelling data story.