如何获得pandas .p​​lot(kind ='kde')的输出？

Question

当我绘制我的熊猫系列的密度分布时我会使用

.plot(kind='kde')

是否有可能获得该图的输出值？如果是的话怎么做？我需要绘制的值.

Answer 1

没有输出值.plot(kind='kde'),它返回一个axes对象.

可以通过图中对象访问原始值_x和_y方法matplotlib.lines.Line2D

In [266]:

ser = pd.Series(np.random.randn(1000))
ax=ser.plot(kind='kde')

In [265]:

ax.get_children() #it is the 3nd object
Out[265]:
[<matplotlib.axis.XAxis at 0x85ea370>,
 <matplotlib.axis.YAxis at 0x8255750>,
 <matplotlib.lines.Line2D at 0x87a5a10>,
 <matplotlib.text.Text at 0x8796f30>,
 <matplotlib.text.Text at 0x87a5850>,
 <matplotlib.text.Text at 0x87a56d0>,
 <matplotlib.patches.Rectangle at 0x87a56f0>,
 <matplotlib.spines.Spine at 0x85ea5d0>,
 <matplotlib.spines.Spine at 0x85eaed0>,
 <matplotlib.spines.Spine at 0x85eab50>,
 <matplotlib.spines.Spine at 0x85ea3b0>]
In [264]:
#get the values
ax.get_children()[2]._x
ax.get_children()[2]._y

Answer 2

你也可以直接调用scipy.stats.gaussian_kde()函数,这就是pandas源代码中发生的事情:

https://github.com/pydata/pandas/blob/master/pandas/tools/plotting.py#L284

函数的文档是:

http://docs.scipy.org/doc/scipy/reference/generated/scipy.stats.gaussian_kde.html