Hexbin Plot
Definition
Hexagonal binning plot
Also known as
Hexagonal binning plot, hexagonal binned plot, hex density plot, hexagonal heatmap, hexagonal histogram
Anatomy
A hexbin plot represents the density of data points in a two-dimensional space by dividing the plane into regular hexagonal cells (or “bins”). Each hexagon represents a region of the 2D space, and its color indicates the number of data points falling within that region. The key components include:
- Hexagonal bins: Regular hexagons that partition the coordinate plane
- Color scale: A sequential color gradient representing the count or density of points in each bin
- Axes: Two continuous variables plotted on the x and y axes
- Legend: A color key indicating how the colors map to frequencies or densities
Interpreting a hexbin plot
Hexbin plots should be read by identifying regions of high and low density. The overall pattern reveals the joint distribution of the two variables and can highlight:
- Clusters and hotspots where data points concentrate
- Sparse regions with few observations
- Unimodality or multimodality in the distribution
- Correlation patterns, linear or nonlinear associations between variables
When and how to use a hexbin plot
Strengths
- Handling of large datasets: Excels with thousands or millions of points where scatter plots would suffer from overplotting
- Reveals density patterns: Shows the true distribution of data better than overplotted scatterplots
- Computational efficiency: More efficient to render than plotting individual points
- Hexagonal advantage: Hexagons have better visual properties than squares, with more uniform distances to neighboring cells
Caveats and limitations
- Bin size sensitivity: Results can vary significantly based on the chosen bin size
- Loss of individual data points: Individual observations are no longer visible
- Learning curve: Less intuitive for audiences unfamiliar with the format
- Color perception issues: Effectiveness depends on a well-chosen color scale
- Small dataset limitations: Not suitable for small datasets where individual points matter
Recommendations
- Choose an appropriate bin size: too large and you lose detail; too small and the pattern becomes fragmented
- Use a sequential color scale that intuitively represents density (e.g., light to dark)
- Include a clear color legend with precise count or density values
- Consider log-transforming the color scale when the density varies dramatically
- Label axes clearly and provide context in the caption
Variations and related visualizations
- 2D histograms: Similar to hexbin plots but using rectangular bins instead of hexagons
- Contour plots: Show density using contour lines rather than filled regions
- Kernel density plots: Using smoothing functions to estimate continuous density
- Bubble charts: Vary the size of points rather than using bins