ggplot2 geom_point() in R: Scatter Plots With Examples

The geom_point() function in ggplot2 draws a scatter plot, mapping x and y values to point positions. You can map additional variables to color, size, shape, or transparency to encode more information per point.

ggplot(df, aes(x, y)) + geom_point()                       # basic scatter
ggplot(df, aes(x, y, color = group)) + geom_point()        # color by group
ggplot(df, aes(x, y, size = z)) + geom_point()             # size by variable
ggplot(df, aes(x, y, shape = group)) + geom_point()        # shape by group
ggplot(df, aes(x, y)) + geom_point(alpha = 0.5)            # transparency
ggplot(df, aes(x, y)) + geom_jitter(width = 0.2)           # jittered points
ggplot(df, aes(x, y)) + geom_point() + facet_wrap(~ group) # one panel per group

What geom_point() does in one sentence

geom_point() maps each row of data to one point on a 2D plane, with x and y positions controlled by aesthetics. Additional aesthetics like color, size, shape, and alpha let you encode extra dimensions: a fifth variable can become point color, a sixth point size, etc.

It is the default tool for showing relationships between two continuous variables, the workhorse of exploratory data analysis, and the foundation for many compound plots (scatter + regression line, scatter + facets, bubble charts).

Syntax

geom_point() is a layer added to a ggplot() call. The minimum is aes(x, y) mapping. Additional aesthetics go either inside aes() (variable-mapped) or outside (constant for all points).

The full signature:

geom_point(mapping = NULL, data = NULL, stat = "identity", position = "identity",
           ..., na.rm = FALSE, show.legend = NA, inherit.aes = TRUE)

Most common usage just passes inherited aesthetics from the parent ggplot() call. Override specific aesthetics for layer-specific control.

Seven common patterns

1. Basic scatter plot

The minimum: ggplot() declares the data and aesthetics; geom_point() adds the layer.

2. Color by group

color = factor(cyl) maps cyl to discrete colors. Wrapping in factor() ensures discrete (categorical) coloring. Without factor(), ggplot treats cyl as continuous and uses a gradient.

3. Size by variable (bubble chart)

Mapping size = hp makes each point's diameter proportional to horsepower. Combined with alpha = 0.6, overlapping points stay visible.

4. Shape by group

Use shape for accessible categorical encoding when colors might not print well or for colorblind-friendly plots.

5. Transparency for overplotting

In dense data (53,940 diamonds), full-opacity points overlap and obscure density. alpha = 0.1 makes individual points faint; clusters appear darker because many overlap.

6. Jittered points for discrete x

geom_jitter() adds a small random offset to each point. Without jitter, points at identical x values would stack on top of each other.

7. Faceted scatter

facet_wrap() creates one panel per unique value of cyl. Each panel has its own x and y axis (default shared). Useful for comparing relationships across groups.

geom_point() vs base R plot()

Base R plot() is one-shot; ggplot2 geom_point() is composable. Base R is faster for quick interactive plots; ggplot2 is more flexible for publication graphics.

When to use which:

• Use ggplot2 for layered, publication-quality, faceted plots.
• Use base R plot() for one-line interactive exploration.

Common pitfalls

Pitfall 1: continuous variable in color. aes(color = cyl) (without factor()) makes ggplot treat cyl as continuous and use a gradient palette, even though cyl has only 3 values. Wrap in factor(cyl) for discrete colors.

Pitfall 2: aesthetic vs constant confusion. geom_point(color = "red") paints all points red. geom_point(aes(color = "red")) creates a fake column "red" and colors by that (resulting in one color, with a confusing legend). The correct constant form has color OUTSIDE aes().

Pitfall 3: jitter changes x values. geom_jitter(width = 0.5) moves each point up to 0.5 units left or right of its true x. For categorical x this is fine; for continuous x where exact values matter, use geom_point with alpha instead.

Try it yourself

Related ggplot2 functions

After mastering geom_point(), look at:

• geom_jitter(): scatter with random offset (for overlapping discrete x)
• geom_smooth(): trend line on top of scatter
• geom_text(), geom_label(): add text annotations to points
• geom_hex(), geom_density_2d(): density alternatives for big data
• scale_color_manual(), scale_color_brewer(): customize point colors
• aes(color, size, shape, alpha): the core aesthetics for scatter plots

For maps with geographic points, geom_sf() works similarly with sf objects. For interactive scatter plots, wrap with plotly::ggplotly().

FAQ

How do I change the color of all points in geom_point?

Use geom_point(color = "blue") (color OUTSIDE aes()). To map color from a column: aes(color = column_name) INSIDE the aes() call.

How do I make points larger or smaller in ggplot2?

For uniform size: geom_point(size = 3). To map size to a variable: geom_point(aes(size = column_name)). Default size is 1.5; values 2 to 4 are typical for clear plots.

How do I deal with overlapping points in geom_point?

Three common fixes. Use alpha = 0.3 for transparency. Use geom_jitter() for small random offsets (best for categorical x). Switch to geom_hex() or geom_density_2d() for very dense data.

Can I use geom_point with a categorical x axis?

Yes, but points stack at each x value. Use geom_jitter() instead, or set position = position_jitter(width = 0.2) inside geom_point() to spread points out at each category.

How do I add a regression line to a scatter plot in ggplot2?

Add geom_smooth(method = "lm") after geom_point(). For LOESS smoothing instead, use geom_smooth(method = "loess") or just geom_smooth() (default). Add se = FALSE to hide the confidence band.