熊猫/ Pyplot直方图：可以绘制DF，但不能绘制子集

Question

df is an enormous dataframe. I only need the subset where Zcoord > 1.

df = pandas.DataFrame(first)
df.columns = ['Xcoord', 'Ycoord', 'Zcoord', 'Angle']
df0 = df[df.Zcoord>1]

The very same code that will draw a histogram of df will not work for df0.

plot1 = plt.figure(1)
plt.hist(df0.Zcoord, bins=100, normed=False)
plt.show()

Ipython spits out KeyError:0.

python 2.7.9 anaconda, ipython 2.2.0, OS 10.9.4

---------------------------------------------------------------------------
KeyError                                  Traceback (most recent call last)
<ipython-input-42-71643df3888f> in <module>()
      1 plot1 = plt.figure(1)
----> 2 plt.hist(df0.Zcoord, bins=100, normed=False)
      3 
      4 plt.show()
      5 from matplotlib.backends.backend_pdf import PdfPages

/Users/Kit/anaconda/lib/python2.7/site-packages/matplotlib/pyplot.pyc in hist(x, bins, range, normed, weights, cumulative, bottom, histtype, align, orientation, rwidth, log, color, label, stacked, hold, **kwargs)
   2888                       histtype=histtype, align=align, orientation=orientation,
   2889                       rwidth=rwidth, log=log, color=color, label=label,
-> 2890                       stacked=stacked, **kwargs)
   2891         draw_if_interactive()
   2892     finally:

/Users/Kit/anaconda/lib/python2.7/site-packages/matplotlib/axes/_axes.pyc in hist(self, x, bins, range, normed, weights, cumulative, bottom, histtype, align, orientation, rwidth, log, color, label, stacked, **kwargs)
   5560         # Massage 'x' for processing.
   5561         # NOTE: Be sure any changes here is also done below to 'weights'
-> 5562         if isinstance(x, np.ndarray) or not iterable(x[0]):
   5563             # TODO: support masked arrays;
   5564             x = np.asarray(x)

/Users/Kit/anaconda/lib/python2.7/site-packages/pandas/core/series.pyc in __getitem__(self, key)
    482     def __getitem__(self, key):
    483         try:
--> 484             result = self.index.get_value(self, key)
    485 
    486             if not np.isscalar(result):

/Users/Kit/anaconda/lib/python2.7/site-packages/pandas/core/index.pyc in get_value(self, series, key)
   1194 
   1195         try:
-> 1196             return self._engine.get_value(s, k)
   1197         except KeyError as e1:
   1198             if len(self) > 0 and self.inferred_type in ['integer','boolean']:

/Users/Kit/anaconda/lib/python2.7/site-packages/pandas/index.so in pandas.index.IndexEngine.get_value (pandas/index.c:2993)()

/Users/Kit/anaconda/lib/python2.7/site-packages/pandas/index.so in pandas.index.IndexEngine.get_value (pandas/index.c:2808)()

/Users/Kit/anaconda/lib/python2.7/site-packages/pandas/index.so in pandas.index.IndexEngine.get_loc (pandas/index.c:3440)()

KeyError: 0

Answer 1

You are passing a pandas.Series to matplotlib ( df0.Zcoord ). However, at the moment, matplotlib is a bit indecisive about whether or not it likes being fed pandas datatypes (as opposed to numpy ndarray 's).

At some point in the bowels of the matplotlib source, the histogram function is probably trying to get the "first item I've been asked to deal with", and it probably does that with a call to input[0] where input is whatever it was asked to chew on. If input is a numpy.ndarray then everything works great. However, if input is a pandas.Series or (even worse) a pandas.DataFrame , the expression input[0] will have a very different meaning. In that case, depending on the structure of the data you fed to plt.hist , there could well be a KeyError when trying to index into your input.

In your particular case, this is probably working fine on df as a whole because df likely has an integer index ( [0, 1, 2, ..., len(df)-1] ), which is the default row index in a DataFrame . However, when you select within df to make df0 , the result keeps winds up with an index that is a subset of the index of df (maybe it winds up [3, 6, 9, 12, ...] ). So everything works fine on df (where the index contains 0 ), but blows chunks on df0 (where, ironically, given its name, 0 does not appear in the index).

Quick fix...instead of

plt.hist(df0.Zcoord, bins=100, normed=False)

run this

plt.hist(df0.Zcoord.values, bins=100, normed=False)

and my guess is everything will be good.