如何用Pandas plot()分组将多条曲线绘制在同一图表中

问题场景 #

现有包含3组店铺销售数据的CSV文件，每组数据对应一条待绘制的曲线。使用Pandas的plot()方法按shop字段分组绘图时，生成了3个独立图表，需要将3条曲线展示在同一个图表中。

原始CSV数据 #

shop,timestamp,sales
north,2023-01-01,235
north,2023-01-02,147
north,2023-01-03,387
north,2023-01-04,367
north,2023-01-05,197
south,2023-01-01,235
south,2023-01-02,98
south,2023-01-03,435
south,2023-01-04,246
south,2023-01-05,273
east,2023-01-01,197
east,2023-01-02,389
east,2023-01-03,87
east,2023-01-04,179
east,2023-01-05,298

原测试代码（生成独立图表） #

import pandas as pd

csv = pd.read_csv('./tmp/sample.csv')
csv.timestamp = pd.to_datetime(csv.timestamp)

# 此代码会生成3个独立图表
csv.plot(x='timestamp', by='shop')

解决方案 #

方法1：数据透视后直接绘图 #

通过pivot()将数据重构为宽表格式，让不同店铺的销售数据对应同一时间戳的列，再调用plot()即可自动将所有曲线绘制在同一图表。

import pandas as pd

csv = pd.read_csv('./tmp/sample.csv')
csv.timestamp = pd.to_datetime(csv.timestamp)

# 透视重构数据：timestamp为索引，shop为列，sales为值
pivoted_df = csv.pivot(index='timestamp', columns='shop', values='sales')
# 绘制合并曲线，自动生成图例
pivoted_df.plot()

方法2：手动指定绘图轴 #

创建matplotlib轴对象，循环遍历分组数据，将每条曲线绘制到同一轴上。

import pandas as pd
import matplotlib.pyplot as plt

csv = pd.read_csv('./tmp/sample.csv')
csv.timestamp = pd.to_datetime(csv.timestamp)

# 创建单个图表轴对象
fig, ax = plt.subplots()

# 按shop分组，逐个绘制曲线到同一轴
for shop_name, group_data in csv.groupby('shop'):
    group_data.plot(x='timestamp', y='sales', ax=ax, label=shop_name)

# 显示图表
plt.show()

内容的提问来源于stack exchange，提问作者Florent Georges