Line up plots between two separate axis with matplotlib-CodePudding

I have the following plot

import matplotlib.pyplot as plt
import pandas as pd
import numpy as np

fig = plt.figure(figsize=(10, 6))
ax_dict = fig.subplot_mosaic(
    [
        ["a", "a", "a", "b"],
        ["a", "a", "a", "b"],
        ["a", "a", "a", "b"],
        ["a", "a", "a", "b"],
    ]
)

rng = np.random.default_rng(1)

df = pd.DataFrame(
    {
        "c1": rng.integers(0, 4, 100),
        "c2": rng.integers(0, 5, 100),
    }
)

ax = ax_dict["a"]
pd.crosstab(df["c1"], df["c2"], normalize="columns").mul(100).round(1).T.plot.barh(
    stacked=True, ax=ax
)


stacked_bar_patch = ax.patches[0]


ax.get_legend().remove()
ax.spines["right"].set_visible(False)
ax.spines["top"].set_visible(False)


ax = ax_dict["b"]
counts = df["c2"].value_counts()
ax.barh(
    counts.index,
    counts,
    color="grey",

    height=stacked_bar_patch.get_height(),
)
_ = ax.set_xticks([])
_ = ax.set_yticks([])
ax.spines["left"].set_visible(False)
ax.spines["right"].set_visible(False)
ax.spines["top"].set_visible(False)
ax.spines["bottom"].set_visible(False)

Which looks as :

I would like to arrange the patches on the right plot (ax_dict['b']) so that they're horizontally aligned with the bars from ax_dict['a'].

Currently they're roughly inline - but the bars are higher / lower on the right than the bars on the left (red circles indicate the "gaps", which wouldn't be present if they were lined up exactly):

My question is - how can i create this plot so that the bars line up exactly ?

CodePudding user response：

This issue can easily be fixed by giving both axes the same y limits, in your example this might be done simply by adding:

ax.set_ylim(-0.5,4.5)

for each subplot, i.e. once before and once after the line ax = ax_dict["b"].

Sidenote: you should consider storing each unique axis within a unique variable to avoid confusion, e.g. naming them ax_a and ax_b.