Histograms¶

Distributions have one theme-specific catch — the property cycle does not reach the outline style — and a couple of habits that make them readable at 89 mm wide.

All data here is synthetic and all axis labels are deliberately generic — this notebook is about how a figure is built and styled, not about any particular measurement.

In [1]:

%matplotlib inline

# Jupyter crops every inline preview with bbox_inches="tight". That defeats the exact column widths
# this package exists to guarantee, and because the themes sit the axes close to the canvas edge, the
# crop slices through the top and right spines. Turn it off so previews match the files jf.save() writes.
%config InlineBackend.print_figure_kwargs = {"bbox_inches": None}

# jf.save() reports what it wrote through the "journalfig" logger instead of printing. Raise that one
# logger -- a blanket logging.basicConfig(level="INFO") also switches on fontTools, which logs a dozen
# subsetting lines per PDF.
import logging

logging.basicConfig(format="%(message)s")
logging.getLogger("journalfig").setLevel(logging.INFO)

import numpy as np
import matplotlib.pyplot as plt

import journalfig as jf

# The themes render previews at 600 dpi, which makes every embedded image in a notebook large. This
# only touches the on-screen preview; jf.save() exports at savefig.dpi regardless.
PREVIEW_DPI = 150


def preview(journal, **kwargs):
    """Apply a theme, then drop the on-screen DPI so this notebook stays light."""
    jf.use(journal, **kwargs)
    plt.rcParams["figure.dpi"] = PREVIEW_DPI


rng = np.random.default_rng(0)
print("journalfig", jf.__version__, "| themes:", jf.JOURNALS)

%matplotlib inline # Jupyter crops every inline preview with bbox_inches="tight". That defeats the exact column widths # this package exists to guarantee, and because the themes sit the axes close to the canvas edge, the # crop slices through the top and right spines. Turn it off so previews match the files jf.save() writes. %config InlineBackend.print_figure_kwargs = {"bbox_inches": None} # jf.save() reports what it wrote through the "journalfig" logger instead of printing. Raise that one # logger -- a blanket logging.basicConfig(level="INFO") also switches on fontTools, which logs a dozen # subsetting lines per PDF. import logging logging.basicConfig(format="%(message)s") logging.getLogger("journalfig").setLevel(logging.INFO) import numpy as np import matplotlib.pyplot as plt import journalfig as jf # The themes render previews at 600 dpi, which makes every embedded image in a notebook large. This # only touches the on-screen preview; jf.save() exports at savefig.dpi regardless. PREVIEW_DPI = 150 def preview(journal, **kwargs): """Apply a theme, then drop the on-screen DPI so this notebook stays light.""" jf.use(journal, **kwargs) plt.rcParams["figure.dpi"] = PREVIEW_DPI rng = np.random.default_rng(0) print("journalfig", jf.__version__, "| themes:", jf.JOURNALS)

journalfig 0.5.1 | themes: ('nature', 'aps', 'elsevier')

A filled histogram¶

Give bars an explicit edge: at a single-column width, adjacent bars of the same colour merge into a solid block without one.

In [2]:

preview("elsevier")
sample = rng.normal(0, 1, 2000)

fig, ax = jf.subplots("elsevier", width="single")
ax.hist(sample, bins=30, edgecolor="white", linewidth=0.4)
ax.set_xlabel("Value (a.u.)")
ax.set_ylabel("Count")

preview("elsevier") sample = rng.normal(0, 1, 2000) fig, ax = jf.subplots("elsevier", width="single") ax.hist(sample, bins=30, edgecolor="white", linewidth=0.4) ax.set_xlabel("Value (a.u.)") ax.set_ylabel("Count")

Out[2]:

Text(0, 0.5, 'Count')

Overlapping distributions¶

Outlines beat fills when distributions overlap. One catch: histtype="step" takes the colour from the property cycle but not the linestyle, so every outline comes out solid and the panel is unreadable in greyscale. Set the linestyle explicitly from jf.LINESTYLE_CYCLE.

In [3]:

preview("aps")
samples = [rng.normal(loc, scale, 1500) for loc, scale in [(-1.0, 0.8), (0.0, 1.0), (1.4, 0.6)]]

fig, axs = jf.subplots("aps", 1, 2, width="double", ratio=0.4)
for i, s in enumerate(samples):
    axs[0].hist(s, bins=30, histtype="step")
    axs[1].hist(s, bins=30, histtype="step", linestyle=jf.LINESTYLE_CYCLE[i], label=f"Series {i + 1}")

axs[0].set_title("cycle only — all solid")
axs[1].set_title("linestyle set explicitly")
for ax in axs:
    ax.set_xlabel("Value (a.u.)")
axs[0].set_ylabel("Count")
axs[1].legend()

preview("aps") samples = [rng.normal(loc, scale, 1500) for loc, scale in [(-1.0, 0.8), (0.0, 1.0), (1.4, 0.6)]] fig, axs = jf.subplots("aps", 1, 2, width="double", ratio=0.4) for i, s in enumerate(samples): axs[0].hist(s, bins=30, histtype="step") axs[1].hist(s, bins=30, histtype="step", linestyle=jf.LINESTYLE_CYCLE[i], label=f"Series {i + 1}") axs[0].set_title("cycle only — all solid") axs[1].set_title("linestyle set explicitly") for ax in axs: ax.set_xlabel("Value (a.u.)") axs[0].set_ylabel("Count") axs[1].legend()

Out[3]:

<matplotlib.legend.Legend at 0x10896cd70>

Filled but overlapping¶

histtype="stepfilled" with alpha works when there are only two or three distributions. Remember that transparency is preserved in PDF, SVG and PNG but flattened in EPS.

In [4]:

preview("nature")
fig, ax = jf.subplots("nature", width="single")
for i, s in enumerate(samples):
    ax.hist(s, bins=30, histtype="stepfilled", alpha=0.45, label=f"Series {i + 1}")
ax.set_xlabel("Value (a.u.)")
ax.set_ylabel("Count")
ax.legend()

preview("nature") fig, ax = jf.subplots("nature", width="single") for i, s in enumerate(samples): ax.hist(s, bins=30, histtype="stepfilled", alpha=0.45, label=f"Series {i + 1}") ax.set_xlabel("Value (a.u.)") ax.set_ylabel("Count") ax.legend()

Out[4]:

<matplotlib.legend.Legend at 0x1138f1810>

Shared bins¶

Comparing distributions requires identical bin edges. Computing them once from the pooled data is the only way to guarantee it.

In [5]:

preview("elsevier")
edges = np.histogram_bin_edges(np.concatenate(samples), bins=28)

fig, ax = jf.subplots("elsevier", width="single")
for i, s in enumerate(samples):
    ax.hist(s, bins=edges, histtype="step", linestyle=jf.LINESTYLE_CYCLE[i], label=f"Series {i + 1}")
ax.set_xlabel("Value (a.u.)")
ax.set_ylabel("Count")
ax.legend()
print("bin width:", round(edges[1] - edges[0], 4))

preview("elsevier") edges = np.histogram_bin_edges(np.concatenate(samples), bins=28) fig, ax = jf.subplots("elsevier", width="single") for i, s in enumerate(samples): ax.hist(s, bins=edges, histtype="step", linestyle=jf.LINESTYLE_CYCLE[i], label=f"Series {i + 1}") ax.set_xlabel("Value (a.u.)") ax.set_ylabel("Count") ax.legend() print("bin width:", round(edges[1] - edges[0], 4))

bin width: 0.2391

Densities, and unequal sample sizes¶

density=True normalises the area to one, which is what makes samples of different sizes comparable.

In [6]:

preview("aps")
small, large = rng.normal(0, 1, 200), rng.normal(0, 1, 4000)

fig, axs = jf.subplots("aps", 1, 2, width="double", ratio=0.4)
for ax, density in zip(axs, [False, True]):
    ax.hist(small, bins=25, histtype="step", density=density, label="n = 200")
    ax.hist(large, bins=25, histtype="step", linestyle="--", density=density, label="n = 4000")
    ax.set_xlabel("Value (a.u.)")
    ax.set_ylabel("Density" if density else "Count")
    ax.set_title("density=True" if density else "counts")
axs[1].legend()

preview("aps") small, large = rng.normal(0, 1, 200), rng.normal(0, 1, 4000) fig, axs = jf.subplots("aps", 1, 2, width="double", ratio=0.4) for ax, density in zip(axs, [False, True]): ax.hist(small, bins=25, histtype="step", density=density, label="n = 200") ax.hist(large, bins=25, histtype="step", linestyle="--", density=density, label="n = 4000") ax.set_xlabel("Value (a.u.)") ax.set_ylabel("Density" if density else "Count") ax.set_title("density=True" if density else "counts") axs[1].legend()

Out[6]:

<matplotlib.legend.Legend at 0x113c35810>

Hatching¶

Hatching distinguishes filled bars without any colour at all. The themes set hatch.linewidth so it stays visible at print size rather than turning into a grey smear.

In [7]:

preview("nature")
fig, ax = jf.subplots("nature", width="single")
for i, (s, hatch) in enumerate(zip(samples[:2], ["///", "\\\\"])):
    ax.hist(s, bins=edges, histtype="stepfilled", facecolor="none", hatch=hatch,
            edgecolor=jf.COLOR_CYCLE[i], label=f"Series {i + 1}")
ax.set_xlabel("Value (a.u.)")
ax.set_ylabel("Count")
ax.legend()

preview("nature") fig, ax = jf.subplots("nature", width="single") for i, (s, hatch) in enumerate(zip(samples[:2], ["///", "\\\\"])): ax.hist(s, bins=edges, histtype="stepfilled", facecolor="none", hatch=hatch, edgecolor=jf.COLOR_CYCLE[i], label=f"Series {i + 1}") ax.set_xlabel("Value (a.u.)") ax.set_ylabel("Count") ax.legend()

Out[7]:

<matplotlib.legend.Legend at 0x113e18690>