I have created a Pandas DataFrame and plotted it using .hist().

我已经创建了一个Pandas DataFrame，并使用.hist()绘制了它。

I want to be able to draw lines/curves on top of the same figure.

我希望能够在同一图形上绘制直线/曲线。

How can I do that?

我怎么能做到这一点？

I was able to plot my data as a histogram using df.hist(column='Example', bins=15). This assigns the returned object to axis.

我能够使用df.hist将数据绘制为直方图(Column=‘Example’，bins=15)。这会将返回的对象分配给AXIS。

I thought I might be able to plot a line using ax=axis as an argument. But this isn't valid.

我想我也许能够使用ax=axis作为参数来绘制一条线。但这是不成立的。

It appears that plt.plot takes different kwargs to DataFrame.hist. Multiple sets of data from a DataFrame can be plot on the same figure, as histograms, using .hist() in combination with an argument of ax=axis.

Plt.plot似乎对DataFrame.hist有不同的看法。通过结合使用.hist()和参数ax=axis，可以将来自DataFrame的多组数据像直方图一样绘制在同一图形上。

Here is some example code, taken from a Jupyter Notebook, plus some data to play with.

以下是取自Jupyter笔记本的一些示例代码，以及一些可供使用的数据。

data = [211995, 139950, 202995, 223000, 184995, 82000, 127000, 240000, 116000, 74500, 151000, 149000, 290000, 146000, 174500, 418000, 150000, 150000, 260000, 100000, 282500, 510000, 142000, 382000, 220000, 259000, 330000, 177500, 290000, 280000, 118000, 97000, 124000, 385000, 199950, 90000, 135000, 395000, 182000, 105000, 80000, 230000, 227950, 176995, 110000, 142000, 132500, 100000, 95000, 257500, 186000, 230000, 169995, 167995, 119950, 119950, 361000, 125000, 242000, 240000, 205000, 187500, 180000, 146000, 257995, 380000, 144995, 139995, 159995, 265000, 288000, 288000, 162500, 290000, 182737, 235000, 250000, 175000, 153000, 125000, 170000, 165000, 187995, 250000, 220000, 108750, 125000, 245000, 100000, 130000, 115000, 218000, 190000, 435000, 300000, 465000, 179950, 259500, 187000, 200000]

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

df = pd.DataFrame(example_data)
df.columns = ['Example']

axis = df.hist(column='Example', bins = 15)

x = numpy.linspace(1e5, 5e5, 20)

def f(x):
    return x * numpy.exp(-x)

y = f(x)
plt.plot(x, y, axis)

• The following explains how to use the object returned by a pandas histogram plot, to plot into a specific axes.


• There are three pandas plotting methods for creating a histogram:
  
  
  
  1. pandas.DataFrame.hist, which returns matplotlib.AxesSubplot or numpy.ndarray of them
  
  
  2. pandas.DataFrame.plot.hist, which returns matplotlib.AxesSubplot
  
  
  3. pandas.DataFrame.plot with kind='hist', which returns matplotlib.axes.Axes or numpy.ndarray of them
  
  
  
  
  
  
  • The pandas plotting API is very versitile, so it's recommended to always check the type of object returned by the plot.
  
  
  
  


• It is better to use the explicit axes interface, than to switch to the implict pyplot interface, as per Why be explicit?
  
  
  
  • matplotlib.pyplot.plot doesn't have an ax parameter, and will only draw onto the current axes.
  
  
  • matplotlib.axes.Axes.plot draws onto the specified axes object.
  
  
  • In the case of the following examples, plt.axhline and Axes.axhline were used for convenience.
  
  
  
  

Data and Imports

import pandas as pd
import seaborn as sns  # for data
import matplotlib.pyplot as plt

df = sns.load_dataset('penguins')

  species     island  bill_length_mm  bill_depth_mm  flipper_length_mm  body_mass_g     sex
0  Adelie  Torgersen            39.1           18.7              181.0       3750.0    Male
1  Adelie  Torgersen            39.5           17.4              186.0       3800.0  Female
2  Adelie  Torgersen            40.3           18.0              195.0       3250.0  Female
3  Adelie  Torgersen             NaN            NaN                NaN          NaN     NaN
4  Adelie  Torgersen            36.7           19.3              193.0       3450.0  Female

df.hist

axes = df.hist(['bill_length_mm', 'bill_depth_mm'])

# draws first axes
axes[0][0].axhline(y=30, c='r')

# draws on second axes
axes[0][1].axhline(y=30, c='purple')

# only drawn an the last plot
plt.axhline(y=20, c='k')

# returns an array of axes
axes
[out]:
array([[<Axes: title={'center': 'bill_length_mm'}>, <Axes: title={'center': 'bill_depth_mm'}>]], dtype=object)

# with only a single column
axes = df.hist('bill_length_mm')

# index to the axes
axes[0][0].axhline(y=30, c='r')

# switching to the implicit interface works, but is not the recommended way
plt.axhline(y=20, c='k')

# still an array with a single axes
axes
[out]:
array([[<Axes: title={'center': 'bill_length_mm'}>]], dtype=object)

df.plot.hist

Single axes

ax = df['bill_length_mm'].plot.hist()

# or
ax = df.plot.hist()

# or
ax = df.plot.hist(y='bill_length_mm')

# plot on the axes
ax.axhline(y=30, c='k')

# returns a single axes
ax
[out]:
<Axes: ylabel='Frequency'>

Array of axes

axes = df.plot.hist(by='sex')

# draws first axes
axes[0].axhline(y=30, c='r')

# draws on second axes
axes[1].axhline(y=30, c='purple')

# only drawn an the last plot
plt.axhline(y=20, c='k')

# returns an array of axes
axes
[out]: 
array([<Axes: title={'center': 'Female'}, ylabel='Frequency'>, <Axes: title={'center': 'Male'}, ylabel='Frequency'>], dtype=object)

df.plot(kind='hist')

Single axes

ax = df['bill_length_mm'].plot(kind='hist')

# or
ax = df.plot(kind='hist')

# or
ax = df.plot(kind='hist', y='bill_length_mm')

# plot on the axes
ax.axhline(y=30, c='k')

# returns a single axes
ax
[out]: 
<Axes: ylabel='Frequency'>

Array of axes

axes = df.plot(kind='hist', by='sex')

# draws first axes
axes[0].axhline(y=30, c='r')

# draws on second axes
axes[1].axhline(y=30, c='purple')

# only drawn an the last plot
plt.axhline(y=20, c='k')

# returns an array of axes
axes
[out]:
array([<Axes: title={'center': 'Female'}, ylabel='Frequency'>, <Axes: title={'center': 'Male'}, ylabel='Frequency'>], dtype=object)

It's just this. Apparently the axis kwarg is not required.

事情就是这样。显然，轴心术并不是必需的。

plt.plot(x, y)
plt.show()

Full example:

完整示例：

df = pd.DataFrame(example_data)
df.columns = ['Example']

axis = df.hist(column='Example', bins = 15)

x = numpy.linspace(1e5, 5e5, 20)

def f(x):
    return 1.0e-5 * x * numpy.exp(-1.0e-5 * x)

y = f(x)
plt.plot(x, y)
plt.show()