Python 创建两个下降气泡图，并根据在同一数据集中找到的值将它们连接起来

我有以下数据，其中我想使用气泡图（较大的值有较大的气泡）在动物和食物之间按降序创建一个并排比较，并用线分别连接它们

数据表格式：

Animal  Food
Cat Hard_Food
Cat Hard_Food
Cat Hard_Food
Cat Hard_Food
Cat Soft_Food
Cat Soft_Food
Cat Soft_Food
Cat Mouse
Cat Soft_Food
Cat Soft_Food
Dog Hard_Food
Dog Hard_Food
Dog Hard_Food
Dog Hard_Food
Dog Soft_Food
Dog Soft_Food
Dog Soft_Food
Dog Soft_Food
Dog Meat
Snake   Mouse
Snake   Meat
Snake   Meat
Snake   Meat

汇总表格式：

    Hard_Food   Meat    Soft_Food   Mouse   Grand Total
Cat        4       0       5           1    10
Dog        4       1       4           0    9
Snake      0       3       0           1    4
GrandTotal 8       4       9           2    23

Python数据帧：

ani_foo = {'Animal': ['Cat','Cat','Cat','Cat','Cat','Cat','Cat','Cat','Cat','Cat','Dog','Dog','Dog','Dog','Dog','Dog','Dog','Dog','Dog','Snake','Snake','Snake','Snake'],
        'Food': ['Hard_Food','Hard_Food','Hard_Food','Hard_Food','Soft_Food','Soft_Food','Soft_Food','Mouse','Soft_Food','Soft_Food','Hard_Food','Hard_Food','Hard_Food','Hard_Food','Soft_Food','Soft_Food','Soft_Food','Soft_Food','Meat','Mouse','Meat','Meat','Meat']
        }

df = pd.DataFrame(ani_foo, columns = ['Animal', 'Food'])

所需输出（通过Excel手动创建）：

---

您可以尝试使用networkx库在两部分网络上创建：

import pandas as pd
import networkx as nx
import matplotlib.pyplot as plt

ani_foo = {'Animal': ['Cat','Cat','Cat','Cat','Cat','Cat','Cat','Cat','Cat','Cat','Dog','Dog','Dog','Dog','Dog','Dog','Dog','Dog','Dog','Snake','Snake','Snake','Snake'],
        'Food': ['Hard_Food','Hard_Food','Hard_Food','Hard_Food','Soft_Food','Soft_Food','Soft_Food','Mouse','Soft_Food','Soft_Food','Hard_Food','Hard_Food','Hard_Food','Hard_Food','Soft_Food','Soft_Food','Soft_Food','Soft_Food','Meat','Mouse','Meat','Meat','Meat']
        }

df = pd.DataFrame(ani_foo, columns = ['Animal', 'Food'])

fig, ax = plt.subplots(figsize=(15,8))
G = nx.from_pandas_edgelist(df, 'Animal', 'Food')
G.add_nodes_from(df['Animal'], bipartite=0)
G.add_nodes_from(df['Food'], bipartite=1)
s = df.stack().value_counts()
s1 = s.index +'\n'+ s.astype(str)
pos = {node:[0, i] for i, node in enumerate(df['Animal'])}
pos.update({node:[1,i] for i, node in enumerate(df['Food'])})
color_dict = {'Cat':'g', 'Dog':'b', 'Snake':'y'}
ec = [color_dict[c] for i in G.edges for c in i if c in color_dict.keys()]
nx.draw_networkx(G, 
                 node_size=[s[i]*250 for i in G.nodes], 
                 pos=pos, 
                 labels = s1.to_dict(), 
                 node_color='lightblue',
                 edge_color=ec)
plt.axis('off')

输出：

---

看起来像是一个非常手动但直接的过程。谢谢分享！有没有办法调整绘图中的图像以避免标签被切断？我尝试了在Stackoverlfow上找到的各种选项，但我无法让它们发挥作用，我认为您可能有不同的方法。例如，如果将“ani_foo”中的第一个“cat”更改为“cattttttttttttttttt”，则“C”在图表中不再可见。