Matplotlib:在单个图上绘制两列或更多列的计数图 [英] Matplotlib: Plot countplot for two or more column on single plot

问题描述

我的数据帧，df:

Sno | Attribute_1   | Attribute_2   | Attribute_3
__________________________________________________
1   | option_1      | option_3      |option_2
2   | option_1      | option_1      |option_1
3   | option_2      | option_2      |option_2
4   | option_1      | option_1      |option_3
5   | option_3      | option_2      |option_2
6   | option_3      | option_3      |option_1
7   | option_1      | option_3      |option_2

此处的 Attribute_1、Attribute_2 和 Attribute_3 包含分类数据 - 每一行的 option_1 或 option_2 或 option_3.

Here Attribute_1, Attribute_2 and Attribute_3 contains categorical data - option_1 or option_2 or option_3 for each of the rows.

我想在同一个图上为所有属性创建一个计数图.我可以通过以下方式在一列中做到这一点:

I want to create a count plot on the same plot for all the attributes. I am able to do it for one column by:

sns.countplot(x="Attribute_1", data=df);

我可以为每个属性单独创建，但我正在寻找的是在同一个图上我可以为所有属性计算图.即X轴将具有属性，每个属性将具有三个计数图.

I can individually create for each of the attributes, but what I am looking for it that on the same plot I can have count plot for all the attributes. i.e X-axis will have attributes, and each attribute will have three count plot.

推荐答案

Seaborn 通常最适合长格式数据集.IE.而不是为每个属性分配不同选项的3列，您将有两列，一列用于选项，一列用于属性.这可以通过 pd.melt 轻松创建.然后可以在选项"上使用 hue 值.列:

Seaborn usually works best with long form datasets. I.e. instead of 3 columns with different options for each attribute you would have two columns, one for the options and one for the attributes. This can easily be created via pd.melt. Then the hue value can be used on the "options" column:

sns.countplot(x="variable", hue="value", data=pd.melt(df))

完整示例:

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

a= np.random.choice(["option_{}".format(i) for i in [1,2,3]], size=(12,3))
df = pd.DataFrame(a, columns=["Attribute_{}".format(i) for i in list("ABC")])

sns.countplot(x="variable", hue="value", data=pd.melt(df))

plt.show()

同样，您可以互换 x 和 hue :

sns.countplot(x="value", hue="variable", data=pd.melt(df))

这篇关于Matplotlib:在单个图上绘制两列或更多列的计数图的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！