KDE - kernel density estimate

seaborn.kdeplot() draws a Kernel Density Estimate (KDE) plot — a smoothed version of a histogram.

Instead of showing discrete bins, it estimates the probability density function (PDF) of a continuous variable.
This makes it perfect for visualising:

• The shape of a distribution
• Comparisons between multiple distributions
• Smoothed trends rather than raw counts

Syntax:

Parameters:

• data = DataFrame containing the data
• x, y = Variables for 1D or 2D density
• hue = Adds separate KDEs for subgroups
• fill = Fill the area under the curve (default False)
• multiple = How multiple hues are displayed ("layer", "stack", "fill")
• common_norm = Whether densities are normalised together or separately
• bw_adjust = Bandwidth adjustment (controls smoothness)
• cut = Extent of curve beyond data range
• gridsize = Number of evaluation points (resolution)
• cmap = Colourmap for 2D plots
• levels = Number of contour levels (for 2D plots)

Basic example

Shows a smooth density curve of the total_bill variable - similar to a histogram, but continuous.

Fill the area under the curve

Fills the area under the KDE curve — great for visual clarity.

Add multiple distributions with hue

Draws one curve per group (Male vs. Female), coloured separately.

Stack or normalise multiple distributions

Options for multiple:

• "layer" = Overlapping curves (default)
• "stack" = Stacked densities
• "fill" = Stacked and normalised to 100% height

Adjust smoothness (bandwidth)

• Smaller bw_adjust → more detail (wigglier curve)
• Larger bw_adjust → smoother (less detailed)

2D KDE plot (Bivariate Distribution)

Displays a contour plot showing where data points are most dense (darker = higher density).

2D KDE with filled contours

Adds filled contours, similar to a topographic heatmap.

Hue in 2D KDE

One filled contour per hue group — useful for comparison.

Clip KDE to data range

Restricts the KDE curve to a specific range.

Cumulative distribution (CDF)

Shows how the cumulative probability increases across values.

Orientation (horizontal)

Flip orientation by using y instead of x.

Control density extent (cut)

Prevents the curve from extending beyond the actual data range.