Seaborn catplot(kind='count') 柱状图转饼图

Question

我在 corona-tracking-apps 上的论文中有以下 df（上面使用了 pd.melt）：

    CTQ-tool    opinion
0   Information and awareness purposes  unacceptable
1   Information and awareness purposes  unacceptable
2   Information and awareness purposes  acceptable
3   Information and awareness purposes  acceptable
4   Information and awareness purposes  unacceptable
... ... ...
2827    Central/Local data storage  NaN
2828    Central/Local data storage  NaN
2829    Central/Local data storage  NaN
2830    Central/Local data storage  NaN
2831    Central/Local data storage  NaN
2832 rows × 2 columns

我正在使用 Seaborn 库制作以下猫图：

代码：

g = sns.catplot("opinion", col="CTQ-tool", col_wrap=4, data=df_original_small, kind="count", height=6.5, aspect=.8)

但是，我不想将它们显示在条形图中，而是希望将它们显示为饼图。 Seaborn.catplot 不允许 something kind='count-pie'。有人知道解决方法吗？

在 TiTo 问题后编辑：

这基本上是我希望看到的所有 8 个条形图发生的情况：

Answer 1

我最终使用 matplotlib 库从头开始构建它：

plt.style.use('seaborn')

IAP = df_original_small['Information and awareness purposes'].value_counts().to_frame().T
QE = df_original_small['Quarantine Enforcement'].value_counts().to_frame().T
CTCR = df_original_small['Contact Tracing and Cross-Referencing'].value_counts().to_frame().T
VPID = df_original_small['Voluntary provision of infection data'].value_counts().to_frame().T
QMA = df_original_small['Quarantine Monitoring App'].value_counts().to_frame().T
QRCode = df_original_small['QR code provided registration tracking'].value_counts().to_frame().T

total = pd.concat([IAP, QE, CTCR, VPID, QMA, QRCode])

fig, ax = plt.subplots(nrows=3, ncols=2)

labels = 'acceptable', 'unacceptable'
colors = ['#008fd5', '#fc4f30']
explode = (0, 0.1)
explode2 = (0.2, 0)

plt.title('Pie chart per CTQ-tool')
plt.tight_layout()

ax[0,0].pie(total.iloc[[0]], startangle=90, colors=colors, wedgeprops={'edgecolor': 'black'}, autopct='%1.f%%', explode=explode, shadow=True)
ax[0,0].set_title('Information and awareness purposes', fontweight='bold')
ax[0,1].pie(total.iloc[[1]],  startangle=90, colors=colors, wedgeprops={'edgecolor': 'black'}, autopct='%1.f%%', explode=explode, shadow=True)
ax[0,1].set_title('Quarantine Enforcement', fontweight='bold')
ax[1,0].pie(total.iloc[[2]],  startangle=90, colors=colors, wedgeprops={'edgecolor': 'black'}, autopct='%1.f%%', explode=explode2, shadow=True)
ax[1,0].set_title('Contact Tracing and Cross-Referencing', fontweight='bold')
ax[1,1].pie(total.iloc[[3]], startangle=90, colors=colors, wedgeprops={'edgecolor': 'black'}, autopct='%1.f%%', explode=explode, shadow=True)
ax[1,1].set_title('Voluntary provision of infection data', fontweight='bold')
ax[2,0].pie(total.iloc[[4]], startangle=90, colors=colors, wedgeprops={'edgecolor': 'black'}, autopct='%1.f%%', explode=explode2, shadow=True)
ax[2,0].set_title('Quarantine Monitoring App', fontweight='bold')
ax[2,1].pie(total.iloc[[5]], startangle=90, colors=colors, wedgeprops={'edgecolor': 'black'}, autopct='%1.f%%', explode=explode, shadow=True)
ax[2,1].set_title('QR code provided registration tracking', fontweight='bold')


fig.suptitle('Public Opinion on CTQ-measures', fontsize=20, y=1.07, fontweight='bold', x=0.37)
fig.set_figheight(10)
fig.set_figwidth(7)
fig.legend(loc='best', labels=labels, fontsize='medium')
fig.tight_layout()

fig.savefig('Opinions_ctq')

plt.show()

Answer 2

如果你想快速完成，也可以试试这个：

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

df = pd.DataFrame({'CTQ-tool':np.random.choice(['a','b','c','d'],50),
                  'opinion':np.random.choice(['acceptable','unacceptable'],50)})

fig, ax = plt.subplots(2,2)
ax = ax.flatten()
tab = pd.crosstab(df['CTQ-tool'],df['opinion'])
for i,cat in enumerate(tab.index):
    tab.loc[cat].plot.pie(ax=ax[i],startangle=90)
    ax[i].set_ylabel('')
    ax[i].set_title(cat, fontweight='bold')

Answer 3

问题是关于使用 python 创建饼图，所以我认为您可以使用另一个可视化库，例如 Plotly，除了作为可视化库之外，Plotly 是一个 interactive 可视化库，因此您的所有图表都将是交互式的！

快速浏览 pie chart documentation。

现在，对于你的问题，我创建了一个小数据集并创建了两个饼图来说明代码的样子。

首先，导入需要的库：

import pandas as pd
import plotly.graph_objects as go
from plotly.subplots import make_subplots
from kaleido.scopes.plotly import PlotlyScope # this will be used to export the chart as static image

玩具数据集：

df = pd.DataFrame(
    {
        "CTQ-tool": [
            "Information and awareness purposes",
            "Information and awareness purposes",
            "Information and awareness purposes",
            "Information and awareness purposes",
            "Information and awareness purposes",
            "Information and awareness purposes",
            "Quarantine Enforcement",
            "Quarantine Enforcement",
            "Quarantine Enforcement",
            "Quarantine Enforcement",
            "Quarantine Enforcement",
            "Quarantine Enforcement",
        ],
        "opinion": [
            "unacceptable",
            "unacceptable",
            "unacceptable",
            "unacceptable",
            "acceptable",
            "unacceptable",
            "acceptable",
            "unacceptable",
            "acceptable",
            "unacceptable",
            "unacceptable",
            "unacceptable",
        ],
    }
)

保存独特的不同工具:

tools = df["CTQ-tool"].unique()

创建的汇总数据：

以下代码将按工具类型和意见类型分组，然后创建一个新列 counts 存储每个工具的每种意见类型的计数。

df_agg = df.groupby(by=["CTQ-tool", "opinion"]).size().reset_index(name="counts")

新数据框 df_agg 将是：

|      | CTQ-tool                           | opinion      | counts |
| ---: | :--------------------------------- | :----------- | -----: |
|    0 | Information and awareness purposes | acceptable   |      1 |
|    1 | Information and awareness purposes | unacceptable |      5 |
|    2 | Quarantine Enforcement             | acceptable   |      2 |
|    3 | Quarantine Enforcement             | unacceptable |      4 |

可视化数据（有趣的部分）：由于这个玩具数据只有两个不同的工具，我创建了一个 sub-plot 的行和两列，但你可以根据需要扩展它 rows/columns。

fig = make_subplots(rows=1, cols=2, specs=[[{"type": "domain"}, {"type": "domain"}]])

然后分别添加每个图表（你可以用for循环来完成）：

fig = make_subplots(rows=1, cols=2, specs=[[{"type": "domain"}, {"type": "domain"}]])

# Information and awareness purposes tool
fig.add_trace(
    go.Pie(
        values=df_agg[df_agg["CTQ-tool"] == tools[0]]["counts"],
        labels=df_agg[df_agg["CTQ-tool"] == tools[0]]["opinion"],
        pull=[0.2, 0.0],
        title=tools[0],
    ),
    1,
    1,
)

# Quarantine Enforcement tool
fig.add_trace(
    go.Pie(
        values=df_agg[df_agg["CTQ-tool"] == tools[1]]["counts"],
        labels=df_agg[df_agg["CTQ-tool"] == tools[1]]["opinion"],
        pull=[0.2, 0.0],
        title=tools[1],
    ),
    1,
    2,
)

更新图表布局：

fig.update_layout(title_text="Public Opinion on CTQ-measures")

fig.show()

最后，导出为静态图像：

现在您已经准备好数据并对其进行了可视化，是时候将其保存为图像了。 Plotly 创作者为此构建了一个工具：Kaleido.

您可以简单地使用它，如下所示：

scope = PlotlyScope()
fig_name = "Public-Opinion-on-CTQ-measures"
with open(f"{fig_name}.png", "wb") as f:
    f.write(scope.transform(fig, "png"))

数字将是：

Seaborn catplot(kind='count') 柱状图转饼图

Seaborn catplot (kind='count') change bar chart to pie chart

python

data-visualization

pandas

pie-chart

seaborn