Layer distributions into a ridgeline plot.

Compare many distributions on one shared scale

A ridgeline plot stacks partially overlapping density curves so changes in location, spread, skew, and modality remain visible across an ordered set of groups. It is especially useful for repeated measurements across time, geography, experimental conditions, or cohorts.

Arrange one sample group per column

Place every comparison group in its own numeric column and observations down the rows. Import CSV, TSV, TXT, JSON, XLSX, or XLS data, select the numeric columns to use, and review the values in the editable table before interpreting the curves.

Choose a meaningful order

Input order is best when the columns already represent a sequence such as months or age bands. Sorting by median or mean can expose gradients and clusters, while alphabetical ordering can help readers find named groups. The chosen order is applied consistently to the live chart and exports.

Understand kernel density estimation

Each ridge is a Gaussian kernel density estimate. Smaller bandwidths reveal more local variation but may create unstable bumps; larger bandwidths smooth the curve and may hide real structure. The default uses Silverman’s rule multiplied by the bandwidth control.

Use overlap without hiding evidence

Overlap creates the compact layered appearance that defines a ridgeline plot. Too much overlap can obscure lower curves, especially with opaque fills. Adjust fill opacity and ridge overlap together, and keep baselines visible when readers need a clear reference for each distribution.

Pick the right height scaling

Equal peak height emphasizes shape independently for each group. A shared density scale preserves absolute density differences. Count scaling gives larger samples more visual weight. Report the scaling choice whenever apparent ridge height influences the comparison.

Show exact summaries when needed

Median and mean markers provide compact center estimates, while the statistics table lists counts, quartiles, ranges, standard deviations, and effective bandwidths. Optional sample rugs reveal the observations behind the smoothed curves.

Export a readable result

PNG and SVG exports preserve the visible groups, current ordering, density settings, markers, colors, grid, labels, and optional transparent background. Use a taller export when many ridges are visible so labels and curve silhouettes remain distinct.

Practical example: compare temperatures by month

Use one column for each month and place daily temperatures below its heading. Ordering the ridges chronologically reveals seasonal movement in center, spread, and shape while retaining the distribution within every month. Use a shared value scale and consistent units so a horizontal shift has the same meaning across all rows.

Ridgelines emphasize pattern, not exact area

Vertical overlap and independent peak scaling can make one group look more prominent even when it has fewer observations. Use the statistics table for exact counts and summaries, disclose the scaling mode, and avoid comparing ridge area unless the chosen normalization supports that interpretation.

Check sensitivity to smoothing

Try several nearby bandwidth settings and show rug marks for smaller samples. Peaks that appear only with very narrow smoothing may be sample noise. When exact quartiles and outliers are more important than layered shape, compare the result with a box plot or violin plot.