MatPlotLib、日期时间和 TypeError：输入类型不支持 ufunc 'isfinite'...

Question

这是一小段代码，它在图表的两条线之间产生一个填充区域：

import matplotlib.pyplot as plt
import numpy as np

x = np.arange(0.0, 2, 0.01)
y1 = np.sin(2 * np.pi * x)
y2 = 1.2 * np.sin(4 * np.pi * x)

fig, ax1 = plt.subplots(1, 1, sharex=True)

# Test support for masked arrays.
ax1.fill_between(x, 0, y1)
ax1.set_ylabel('between y1 and 0')
y2 = np.ma.masked_greater(y2, 1.0)
ax1.plot(x, y1, x, y2, color='black')
ax1.fill_between(
    x, y1, y2, where=y2 >= y1,
    facecolor='green',
    interpolate=True)
ax1.fill_between(x, y1, y2, where=y2 <= y1, facecolor='red', interpolate=True)
ax1.set_title('Now regions with y2>1 are masked')

# Show the plot.
plt.show()

看起来像这样：

现在，更改开始以使 x 现在是日期时间对象的集合，如下所示：

import datetime

x1 = np.arange(0.0, 2, 0.01)
now = np.datetime64(datetime.datetime.now())
x = np.array([now - np.timedelta64(datetime.timedelta(seconds=i)) for i in range(200)])
y1 = np.sin(2 * np.pi * x1)
y2 = 1.2 * np.sin(4 * np.pi * x1)

产量：

Traceback (most recent call last):                                                File "fill_between_demo.py", line 21, in <module>                             
    ax1.fill_between(x, 0, y1)                                                  
  File "/home/usr/.virtualenvs/raiju/lib/python3.6/site-packages/matplotlib/__init__.py", line 1898, in inner                                                  
    return func(ax, *args, **kwargs)                                            
  File "/home/usr/.virtualenvs/raiju/lib/python3.6/site-packages/matplotlib/axes/_axes.py", line 4778, in fill_between                                         
    x = ma.masked_invalid(self.convert_xunits(x))                               
  File "/home/usr/.virtualenvs/raiju/lib/python3.6/site-packages/numpy/ma/core.py", line 2388, in masked_invalid                                               
    condition = ~(np.isfinite(a))                                               
TypeError: ufunc 'isfinite' not supported for the input types, and the inputs could not be safely coerced to any supported types according to the casting rule ''safe''     

为什么会发生这种情况以及如何解决？

请注意，绘制数据（也就是不使用fill*）可以正常工作。

Answer 1

问题是，numpy ufunc isfinite 没有为 numpy.datetime64 dtype 定义。不过，我们正在努力改变这一点。此issue on numpy's github 正在此pull-request 中处理，但只要未完成并合并，您将无法在该dtype 上使用isfinite。这是一个问题，因为 matplotlib.pyplot.fill_between 在调用 numpy.ma.masked_invalid 以屏蔽输入数组的所有无效条目时隐式使用此函数。

不过有一个变通方法。正如在answer 中指出的关于fill_between 绘制datetime64 类型的pandas Series 的类似问题，pandas 使用matplotlib 为datetime64 dtype 的（以及其他）numpy 数组注册了一个自定义转换器。要使用它，您只需导入 pandas：

import numpy as np
import matplotlib.pyplot as plt
import datetime
# import pandas for its converter that is then used in pyplot!
import pandas

x1 = np.arange(0.0, 2, 0.01)
now = np.datetime64(datetime.datetime.now())
x = np.array([now - np.timedelta64(datetime.timedelta(seconds=i))
              for i in range(200)])
y1 = np.sin(2 * np.pi * x1)
y2 = 1.2 * np.sin(4 * np.pi * x1)

fig, ax1 = plt.subplots(1, 1, sharex=True)

# Test support for masked arrays.
ax1.fill_between(x, 0, y1)
ax1.set_ylabel('between y1 and 0')
y2 = np.ma.masked_greater(y2, 1.0)
ax1.plot(x, y1, x, y2, color='black')
ax1.fill_between(
    x, y1, y2, where=y2 >= y1,
    facecolor='green',
    interpolate=True)
ax1.fill_between(x, y1, y2, where=y2 <= y1, facecolor='red', interpolate=True)
ax1.set_title('Now regions with y2>1 are masked')

# Show the plot.
plt.show()

将起作用并为您提供所需的输出：

Answer 2

我在使用 ax.setxlims 时遇到了类似的问题，因为我忘记将一些日期时间转换为 np.datetime。

>>> import datetime
>>> import numpy
>>> a = datetime.datetime.now()
>>> numpy.isfinite(a)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: ufunc 'isfinite' not supported for the input types, and the inputs could not be safely coerced to any supported types according to the casting rule ''safe''
>>> a = numpy.datetime64(a)
>>> numpy.isfinite(a)
True

看来不再需要导入pandas了