geom_smooth

Aids the eye in seeing patterns in the presence of overplotting by adding smoothed conditional means. geom_smooth() fits a model to the data and displays the fitted line along with a confidence interval (for linear models). This is particularly useful when you have many data points and want to visualize the underlying trend.

Arguments

• plot::GGPlot (optional): a plot object to add this geom to. This is typically used to facilitate creating your ggplot as part of a @chain.
• data (DataFrame): Data to use for this geom. If not provided, the geom will inherit the data from ggplot.
• aes(...): the names of the columns in the DataFrame that will be used in the mapping
• inherit_aes: should the geom inherit aes from the ggplot?
• ...: options that are not mapped to a column (passed to Makie.Lines)

Smoothing Methods

TidierPlots supports two smoothing methods:

• method="smooth" (default): Uses LOESS (locally estimated scatterplot smoothing), which fits local polynomial regressions. This is flexible and follows the data closely.
• method="lm": Uses linear regression, fitting a straight line through the data. This also displays a 95% confidence band around the line.

Required Aesthetics

• x
• y

Optional Aesthetics (see aes)

• color / colour
• group

Optional Arguments

• method: "smooth" (default, LOESS) or "lm" (linear regression)
• color / colour: line color
• linewidth: line thickness
• alpha: transparency
• linestyle / linetype: line pattern

using TidierPlots
using DataFrames

penguins = penguins_data();

Basic LOESS Smoothing

The default method is LOESS (locally estimated scatterplot smoothing), which creates a flexible smooth line that follows the data:

ggplot(penguins, @aes(x = bill_length_mm, y = bill_depth_mm)) +
    geom_point(alpha = 0.3) +
    geom_smooth()

Combine smooth line with points for context:

xs = range(0, 2pi, length = 50)
ys = sin.(xs) .+ randn(length(xs)) * 0.3
df = DataFrame(x = xs, y = ys)

ggplot(df, @aes(x = x, y = y)) +
    geom_point() +
    geom_smooth()

Linear Regression

Use method="lm" for linear regression. This displays both the fitted line and a 95% confidence band:

ggplot(penguins, @aes(x = bill_length_mm, y = bill_depth_mm)) +
    geom_point(alpha = 0.3) +
    geom_smooth(method = "lm")

Customizing Appearance

Change the color and line width:

ggplot(penguins, @aes(x = bill_length_mm, y = bill_depth_mm)) +
    geom_point(alpha = 0.3) +
    geom_smooth(color = :red, linewidth = 3, alpha = 0.7)

Use line styles:

ggplot(penguins, @aes(x = bill_length_mm, y = bill_depth_mm)) +
    geom_point(alpha = 0.3) +
    geom_smooth(linestyle = :dash, linewidth = 2)

Grouping

The smooth line respects the color aesthetic, fitting separate curves for each group:

ggplot(penguins, @aes(x = bill_length_mm, y = bill_depth_mm, color = species)) +
    geom_point(alpha = 0.3) +
    geom_smooth()

With linear regression by group:

ggplot(penguins, @aes(x = bill_length_mm, y = bill_depth_mm, color = species)) +
    geom_point(alpha = 0.3) +
    geom_smooth(method = "lm")

Comparing Methods

Compare LOESS and linear regression on the same plot:

ggplot(penguins, @aes(x = bill_length_mm, y = bill_depth_mm)) +
    geom_point(alpha = 0.2) +
    geom_smooth(color = :blue) +
    geom_smooth(method = "lm", color = :red)

Handling Noisy Data

LOESS smoothing is particularly useful for visualizing trends in noisy data:

noisy_df = DataFrame(
    x = 1:100,
    y = sin.((1:100) ./ 10) .+ randn(100) .* 0.5
)

ggplot(noisy_df, @aes(x = x, y = y)) +
    geom_point(alpha = 0.5) +
    geom_smooth(color = :red, linewidth = 2)

---

This page was generated using Literate.jl.