Question

我试图从像这样的数据框中制作一系列折线图

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

df = pd.DataFrame({ 'CITY' : np.random.choice(['PHOENIX','ATLANTA','CHICAGO', 'MIAMI', 'DENVER'], 10000),
                    'DAY': np.random.choice(['Monday','Tuesday','Wednesday', 'Thursday', 'Friday', 'Saturday', 'Sunday'], 10000),
                    'TIME_BIN': np.random.randint(1, 86400, size=10000),
                    'COUNT': np.random.randint(1, 700, size=10000)})

df['TIME_BIN'] = pd.to_datetime(df['TIME_BIN'], unit='s').dt.round('10min').dt.strftime('%H:%M:%S')
print(df)

         CITY  COUNT        DAY  TIME_BIN
0     ATLANTA    270  Wednesday  10:50:00
1     CHICAGO    375  Wednesday  12:20:00
2       MIAMI    490   Thursday  11:30:00
3       MIAMI    571     Sunday  23:30:00
4      DENVER    379   Saturday  07:30:00
...       ...    ...        ...       ...
9995  ATLANTA    107   Saturday  21:10:00
9996   DENVER    127    Tuesday  15:00:00
9997   DENVER    330     Friday  06:20:00
9998  PHOENIX    379   Saturday  19:50:00
9999  CHICAGO    628   Saturday  01:30:00

这就是我现在所拥有的：

piv = df.pivot(columns="DAY").plot(x='TIME_BIN', kind="Line", subplots=True)
plt.show()

但是x轴格式混乱，我需要每个城市都是自己的线。我该如何解决这个问题？我认为我需要遍历一周中的每一天而不是尝试在一行中创建一个数组。我没有运气就尝试过seaborn。总而言之，这就是我想要实现的目标：

x轴上的TIME_BIN
Y轴上的COUNT
每个CITY的不同颜色线
每天一张图表

Answer 1

我不知道旋转在这里有多大帮助，因为最后你需要将数据分成两次，一次是针对一周中的几天，这些天数应该分成几个子图，再次针对城市，应有自己的彩色线条。在这一点上，我们处于大熊猫可以用它的绘图包装器做的极限。

Matplotlib

使用matplotlib可以遍历两个类别，天和城市，只绘制数据。

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.dates

df = pd.DataFrame({ 
    'CITY' : np.random.choice(['PHOENIX','ATLANTA','CHICAGO', 'MIAMI', 'DENVER'], 10000),
    'DAY': np.random.choice(['Monday','Tuesday','Wednesday', 'Thursday', 
                             'Friday', 'Saturday', 'Sunday'], 10000),
    'TIME_BIN': np.random.randint(1, 86400, size=10000),
    'COUNT': np.random.randint(1, 700, size=10000)})

df['TIME_BIN'] = pd.to_datetime(df['TIME_BIN'], unit='s').dt.round('10min')


days = ['Monday','Tuesday','Wednesday', 'Thursday', 'Friday', 'Saturday', 'Sunday']
cities = np.unique(df["CITY"])
fig, axes = plt.subplots(nrows=len(days), figsize=(13,8), sharex=True)

# loop over days (one could use groupby here, but that would lead to days unsorted)
for i, day in enumerate(days):
    ddf = df[df["DAY"] == day].sort_values("TIME_BIN")
    # loop over cities
    for city in cities:
        dddf = ddf[ddf["CITY"] == city]
        axes[i].plot(dddf["TIME_BIN"], dddf["COUNT"], label=city)
    axes[i].margins(x=0)
    axes[i].set_title(day)


fmt = matplotlib.dates.DateFormatter("%H:%M") 
axes[-1].xaxis.set_major_formatter(fmt)   
axes[0].legend(bbox_to_anchor=(1.02,1))
fig.subplots_adjust(left=0.05,bottom=0.05, top=0.95,right=0.85, hspace=0.8)    
plt.show()

Seaborn

使用Seaborn FacetGrid可以获得大致相同的效果。

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

df = pd.DataFrame({ 
    'CITY' : np.random.choice(['PHOENIX','ATLANTA','CHICAGO', 'MIAMI', 'DENVER'], 10000),
    'DAY': np.random.choice(['Monday','Tuesday','Wednesday', 'Thursday', 
                             'Friday', 'Saturday', 'Sunday'], 10000),
    'TIME_BIN': np.random.randint(1, 86400, size=10000),
    'COUNT': np.random.randint(1, 700, size=10000)})

df['TIME_BIN'] = pd.to_datetime(df['TIME_BIN'], unit='s').dt.round('10min')

days = ['Monday','Tuesday','Wednesday', 'Thursday', 'Friday', 'Saturday', 'Sunday']
cities = np.unique(df["CITY"])

g = sns.FacetGrid(data=df.sort_values('TIME_BIN'), 
                  row="DAY", row_order=days, 
                  hue="CITY", hue_order=cities, sharex=True, aspect=5)
g.map(plt.plot, "TIME_BIN", "COUNT")

g.add_legend()
g.fig.subplots_adjust(left=0.05,bottom=0.05, top=0.95,hspace=0.8)
fmt = matplotlib.dates.DateFormatter("%H:%M")
g.axes[-1,-1].xaxis.set_major_formatter(fmt)
plt.show()

如何从Pandas数据框中绘制多个折线图

1 个答案:

Matplotlib

Seaborn