ridge

Distributions

Minimal call

import random
rng = random.Random(15)
data = {"row": [], "value": []}
for i, row in enumerate(["a", "b", "c", "d"]):
    data["row"] += [row] * 150
    data["value"] += [rng.gauss(i, 1.2) for _ in range(150)]

c = pt.chart(data, aes(x="row", y="value"), xlabel="value")
c.add_ridge(overlap=1.5)
c.yticks([])

Worked example

import random

import plotlet as pt
from plotlet import aes

rng = random.Random(15)
months = ["Jan", "Feb", "Mar", "Apr", "May"]
data = {"month": [], "temp": []}
for i, month in enumerate(months):
    for _ in range(200):
        data["month"].append(month)
        data["temp"].append(rng.gauss(20 + i, 3))

c = pt.chart(data, aes(x="month", y="temp"),
             title="daily temperature", xlabel="temperature (°C)",
             data_width=320, data_height=240)
c.add_ridge(overlap=1.6)
c.yticks([])

Docstring

Ridgeline / joyplot — stacked, vertically-offset KDE curves.

Seaborn's deprecated `kdeplot(..., multiple="stack")` and R's
`ggridges::geom_density_ridges`. Each series gets its own baseline on y;
densities are KDE-shaped via a Gaussian kernel. Handy for showing
distributions across many groups.

The whole stack is rendered as one artist so it owns its own y-baselines.
Categories are placed at integer y; densities are scaled to a fraction of
the row spacing.

API (long-form only):
  c.add_ridge(aes(x="label", y="value"))
  c.add_ridge(aes(x="label", y="value", color="cohort"))  # overlaid
                                                          # sub-densities
                                                          # per row

Aesthetics:
  color=         bare → literal fill color; aes(color="col") → one
                 overlaid KDE per level within each row (ggridges
                 fill= second factor)
  palette=       maps levels → colors when color is mapped in aes

Styling kwargs:
  overlap=1.4    height of each ridge as a fraction of row spacing
                 (>1 lets neighbouring ridges overlap)
  bw=None        bandwidth override; defaults to Silverman's rule
  n_grid=200     KDE evaluation grid resolution
  alpha=0.6      fill opacity