当初始df包含日期时间序列时，Pandas应用返回的数据帧

Question

当应用一个在pandas数据帧上返回元组的函数时，我遇到了一些我不理解的行为。我的目的是让df.apply()返回一个新的序列，但只有当我将数据帧中的列子集化以排除一个日期时间序列的列时，这种方法才有效。在

这个虚拟示例演示了我看到的行为：

df = pd.DataFrame(np.random.randn(5, 4), columns=list('ABCD'))

def random(row):
    # Return an tuple with more elements than df has columns
    return (1,2,3,4,5,6,7,8)

df.apply(random,axis=1)

# Output, returns new series as expected:
0    (1, 2, 3, 4, 5, 6, 7, 8)
1    (1, 2, 3, 4, 5, 6, 7, 8)
2    (1, 2, 3, 4, 5, 6, 7, 8)
3    (1, 2, 3, 4, 5, 6, 7, 8)
4    (1, 2, 3, 4, 5, 6, 7, 8)

正如预期的那样，但是当我向dataframe添加一个datetime列时。。。在

^{pr2}$

我得到这个错误：

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
/Users/jguillette/anaconda/lib/python3.5/site-packages/pandas/core/internals.py in create_block_manager_from_arrays(arrays, names, axes)
   4262         blocks = form_blocks(arrays, names, axes)
-> 4263         mgr = BlockManager(blocks, axes)
   4264         mgr._consolidate_inplace()

/Users/jguillette/anaconda/lib/python3.5/site-packages/pandas/core/internals.py in __init__(self, blocks, axes, do_integrity_check, fastpath)
   2760         if do_integrity_check:
-> 2761             self._verify_integrity()
   2762 

/Users/jguillette/anaconda/lib/python3.5/site-packages/pandas/core/internals.py in _verify_integrity(self)
   2970             if block._verify_integrity and block.shape[1:] != mgr_shape[1:]:
-> 2971                 construction_error(tot_items, block.shape[1:], self.axes)
   2972         if len(self.items) != tot_items:

/Users/jguillette/anaconda/lib/python3.5/site-packages/pandas/core/internals.py in construction_error(tot_items, block_shape, axes, e)
   4232     raise ValueError("Shape of passed values is {0}, indices imply {1}".format(
-> 4233         passed, implied))
   4234 

ValueError: Shape of passed values is (5, 8), indices imply (5, 5)

During handling of the above exception, another exception occurred:

ValueError                                Traceback (most recent call last)
<ipython-input-29-b57dd4b93995> in <module>()
----> 1 df.apply(random,axis=1)

/Users/jguillette/anaconda/lib/python3.5/site-packages/pandas/core/frame.py in apply(self, func, axis, broadcast, raw, reduce, args, **kwds)
   4150                     if reduce is None:
   4151                         reduce = True
-> 4152                     return self._apply_standard(f, axis, reduce=reduce)
   4153             else:
   4154                 return self._apply_broadcast(f, axis)

/Users/jguillette/anaconda/lib/python3.5/site-packages/pandas/core/frame.py in _apply_standard(self, func, axis, ignore_failures, reduce)
   4263                 index = None
   4264 
-> 4265             result = self._constructor(data=results, index=index)
   4266             result.columns = res_index
   4267 

/Users/jguillette/anaconda/lib/python3.5/site-packages/pandas/core/frame.py in __init__(self, data, index, columns, dtype, copy)
    264                                  dtype=dtype, copy=copy)
    265         elif isinstance(data, dict):
--> 266             mgr = self._init_dict(data, index, columns, dtype=dtype)
    267         elif isinstance(data, ma.MaskedArray):
    268             import numpy.ma.mrecords as mrecords

/Users/jguillette/anaconda/lib/python3.5/site-packages/pandas/core/frame.py in _init_dict(self, data, index, columns, dtype)
    400             arrays = [data[k] for k in keys]
    401 
--> 402         return _arrays_to_mgr(arrays, data_names, index, columns, dtype=dtype)
    403 
    404     def _init_ndarray(self, values, index, columns, dtype=None, copy=False):

/Users/jguillette/anaconda/lib/python3.5/site-packages/pandas/core/frame.py in _arrays_to_mgr(arrays, arr_names, index, columns, dtype)
   5406     axes = [_ensure_index(columns), _ensure_index(index)]
   5407 
-> 5408     return create_block_manager_from_arrays(arrays, arr_names, axes)
   5409 
   5410 

/Users/jguillette/anaconda/lib/python3.5/site-packages/pandas/core/internals.py in create_block_manager_from_arrays(arrays, names, axes)
   4265         return mgr
   4266     except ValueError as e:
-> 4267         construction_error(len(arrays), arrays[0].shape, axes, e)
   4268 
   4269 

/Users/jguillette/anaconda/lib/python3.5/site-packages/pandas/core/internals.py in construction_error(tot_items, block_shape, axes, e)
   4231         raise ValueError("Empty data passed with indices specified.")
   4232     raise ValueError("Shape of passed values is {0}, indices imply {1}".format(
-> 4233         passed, implied))
   4234 
   4235 

ValueError: Shape of passed values is (5, 8), indices imply (5, 5)

我唯一没有得到错误的时候是当函数返回一个元素数与dataframe的列数相同的元组时，当它返回一个dataframe而不是一个序列时。在

有没有办法改变这种行为？在我的例子中，我不需要在函数中使用日期时间信息，但是我仍然不明白排除它如何改变apply的行为。在

任何有见识的人都将不胜感激。在

Answer 1

https://github.com/pandas-dev/pandas/blob/v0.22.0/pandas/core/frame.py#L236-L6142

class DataFrame(NDFrame):
def _apply_standard(self, func, axis, ignore_failures=False, reduce=True):
    # skip if we are mixed datelike and trying reduce across axes
    # GH6125
    # this block will judge if it's datelike mixed type and disable reduce if true
    if (reduce and axis == 1 and self._is_mixed_type and
            self._is_datelike_mixed_type):
        reduce = False
    # try to reduce first (by default)
    # this only matters if the reduction in values is of different dtype
    # e.g. if we want to apply to a SparseFrame, then can't directly reduce
    if reduce:
        values = self.values

        # we cannot reduce using non-numpy dtypes,
        # as demonstrated in gh-12244
        if not is_extension_type(values):
            # Create a dummy Series from an empty array
            index = self._get_axis(axis)
            empty_arr = np.empty(len(index), dtype=values.dtype)
            dummy = Series(empty_arr, index=self._get_axis(axis),
                           dtype=values.dtype)

            try:
                labels = self._get_agg_axis(axis)
                result = lib.reduce(values, func, axis=axis, dummy=dummy,
                                    labels=labels)
                return Series(result, index=labels)
            except Exception:
                pass

    dtype = object if self._is_mixed_type else None
    if axis == 0:
        series_gen = (self._ixs(i, axis=1)
                      for i in range(len(self.columns)))
        res_index = self.columns
        res_columns = self.index
    elif axis == 1:
        res_index = self.index
        res_columns = self.columns
        values = self.values
        series_gen = (Series.from_array(arr, index=res_columns, name=name,
                                        dtype=dtype)
                      for i, (arr, name) in enumerate(zip(values,
                                                          res_index)))
    else:  # pragma : no cover
        raise AssertionError('Axis must be 0 or 1, got %s' % str(axis))

    i = None
    keys = []
    results = {}
    if ignore_failures:
        successes = []
        for i, v in enumerate(series_gen):
            try:
                results[i] = func(v)
                keys.append(v.name)
                successes.append(i)
            except Exception:
                pass
        # so will work with MultiIndex
        if len(successes) < len(res_index):
            res_index = res_index.take(successes)
    else:
        try:
            for i, v in enumerate(series_gen):
                results[i] = func(v)
                keys.append(v.name)
        except Exception as e:
            if hasattr(e, 'args'):
                # make sure i is defined
                if i is not None:
                    k = res_index[i]
                    e.args = e.args + ('occurred at index %s' %
                                       pprint_thing(k), )
            raise

    if len(results) > 0 and is_sequence(results[0]):
        if not isinstance(results[0], Series):
            index = res_columns
        else:
            index = None

        result = self._constructor(data=results, index=index)
        result.columns = res_index

        if axis == 1:
            result = result.T
        result = result._convert(datetime=True, timedelta=True, copy=False)

    else:

        result = Series(results)
        result.index = res_index

    return result

我找到了你问题背后的源代码，详细原因你可以检查GH6125，正如评论所说。 我的决定有些愚蠢，如下所示。在

^{pr2}$

第二个解决方案是确保func返回一个序列（看起来比较慢）

def func(line):
    return pd.Series(['a', 'x', 'y'])
df.apply(func, axis=1)

希望有帮助。在

Answer 2

根据df的数据类型，pandas处理apply返回值的方式不同。在第一个示例中，所有数据类型都是float，而在添加列E之后，数据类型是混合的，这导致pandas试图使用返回的值重建数据帧。我不知道这种行为背后的理性，但以下几点应该可以解决你的问题：

df.astype(object).apply(random,axis=1)
Out[64]: 
0    (1, 2, 3, 4, 5, 6, 7, 8)
1    (1, 2, 3, 4, 5, 6, 7, 8)
2    (1, 2, 3, 4, 5, 6, 7, 8)
3    (1, 2, 3, 4, 5, 6, 7, 8)
4    (1, 2, 3, 4, 5, 6, 7, 8)