如何使用条形 plot 缩放 Seaborn 的 y 轴

Question

I'm using factorplot(kind="bar") .

How do I scale the y-axis, for example with log-scale?

I tried tinkering with the plots' axes, but that always messed up the bar plot in one way or another, so please try your solution first to make sure it really works.

Answer 1

You can use Matplotlib commands after calling factorplot . For example:

import seaborn as sns
import matplotlib.pyplot as plt
sns.set(style="whitegrid")

titanic = sns.load_dataset("titanic")

g = sns.factorplot("class", "survived", "sex",
                   data=titanic, kind="bar",
                   size=6, palette="muted", legend=False)
g.fig.get_axes()[0].set_yscale('log')
plt.show()

Answer 2

Considering your question mentions barplot I thought I would add in a solution for that type of plot also as it differs from the factorplot in @Jules solution.

import seaborn as sns
import matplotlib.pyplot as plt
sns.set(style="whitegrid")

titanic = sns.load_dataset("titanic")

g = sns.barplot(x="class", y="survived", hue="sex",
                data=titanic, palette="muted")
g.set_yscale("log")

Answer 3

If you are facing the problem of vanishing bars upon setting log-scale using the previous solutions, try adding log=True to the seaborn function call instead. (I'm lacking reputation to comment on the other answers).

Using sns.factorplot :

import seaborn as sns
import matplotlib.pyplot as plt
sns.set(style="whitegrid")

titanic = sns.load_dataset("titanic")

g = sns.factorplot(x="class", y="survived", hue="sex", kind='bar',
                   data=titanic, palette="muted", log=True)
g.ax.set_ylim(0.05, 1)

Using sns.barplot :

import seaborn as sns
import matplotlib.pyplot as plt
sns.set(style="whitegrid")

titanic = sns.load_dataset("titanic")

g = sns.barplot(x="class", y="survived", hue="sex",
                data=titanic, palette="muted", log=True)
g.set_ylim(0.05, 1)

Answer 4

Seaborn's catplot does not have anymore the log parameter.

For those looking for an updated answer , here's the quickest fix I've used: you have to use matplotlib's built-in support by accessing the axes object.

g = sns.catplot(data=df, <YOUR PARAMETERS>)
for ax in g.fig.axes:
    ax.set_yscale('log')