Python pandas 多索引列

Python pandas Multindex column

首先,我在 jupyter notebook 中使用 python 3.50。

我想创建一个 DataFrame 来在报表中显示一些数据。我希望它有两个索引列(如果引用它的术语不正确,请原谅。我不习惯使用 pandas)。

我有这个有效的示例代码:

frame = pd.DataFrame(np.arange(12).reshape(( 4, 3)), 
                  index =[['a', 'a', 'b', 'b'], [1, 2, 1, 2]], 
                  columns =[['Ohio', 'Ohio', 'Ohio'], ['Green', 'Red', 'Green']])

但是当我尝试将其用于我的案例时,它给了我一个错误:

cell_rise_Inv= pd.DataFrame([[0.00483211, 0.00511619, 0.00891821, 0.0449637, 0.205753], 
                             [0.00520049, 0.00561577, 0.010993, 0.0468998, 0.207461],
                             [0.00357213, 0.00429087, 0.0132186, 0.0536389, 0.21384],
                             [-0.0021868, -0.0011312, 0.0120546, 0.0647213, 0.224749],
                             [-0.0725403, -0.0700884, -0.0382486, 0.0899121, 0.313639]], 
                            index =[['transition [ns]','transition [ns]','transition [ns]','transition [ns]','transition [ns]'],
                                   [0.0005, 0.001, 0.01, 0.1, 0.5]],
                            columns =[[0.01, 0.02, 0.05, 0.1, 0.5],['capacitance [pF]','capacitance [pF]','capacitance [pF]','capacitance [pF]','capacitance [pF]']])
cell_rise_Inv

---------------------------------------------------------------------------
AssertionError                            Traceback (most recent call last)
<ipython-input-89-180a1ad88403> in <module>()
      6                             index =[['transition [ns]','transition [ns]','transition [ns]','transition [ns]','transition [ns]'],
      7                                    [0.0005, 0.001, 0.01, 0.1, 0.5]],
----> 8                             columns =[[0.01, 0.02, 0.05, 0.1, 0.5],['capacitance [pF]','capacitance [pF]','capacitance [pF]','capacitance [pF]','capacitance [pF]']])
      9 cell_rise_Inv

C:\Users\Josele\Anaconda3\lib\site-packages\pandas\core\frame.py in __init__(self, data, index, columns, dtype, copy)
    261                     if com.is_named_tuple(data[0]) and columns is None:
    262                         columns = data[0]._fields
--> 263                     arrays, columns = _to_arrays(data, columns, dtype=dtype)
    264                     columns = _ensure_index(columns)
    265 

C:\Users\Josele\Anaconda3\lib\site-packages\pandas\core\frame.py in _to_arrays(data, columns, coerce_float, dtype)
   5350     if isinstance(data[0], (list, tuple)):
   5351         return _list_to_arrays(data, columns, coerce_float=coerce_float,
-> 5352                                dtype=dtype)
   5353     elif isinstance(data[0], collections.Mapping):
   5354         return _list_of_dict_to_arrays(data, columns,

C:\Users\Josele\Anaconda3\lib\site-packages\pandas\core\frame.py in _list_to_arrays(data, columns, coerce_float, dtype)
   5429         content = list(lib.to_object_array(data).T)
   5430     return _convert_object_array(content, columns, dtype=dtype,
-> 5431                                  coerce_float=coerce_float)
   5432 
   5433 

C:\Users\Josele\Anaconda3\lib\site-packages\pandas\core\frame.py in _convert_object_array(content, columns, coerce_float, dtype)
   5487             # caller's responsibility to check for this...
   5488             raise AssertionError('%d columns passed, passed data had %s '
-> 5489                                  'columns' % (len(columns), len(content)))
   5490 
   5491     # provide soft conversion of object dtypes

AssertionError: 2 columns passed, passed data had 5 columns

有什么想法吗?我不明白为什么这个例子有效而我的不这样做。 :S

提前谢谢你:)。

看起来确实不一致。我会使用 pd.MultiIndex 构造函数 from_arrays

idx = pd.MultiIndex.from_arrays([['transition [ns]'] * 5,
                                 [0.0005, 0.001, 0.01, 0.1, 0.5]])
col = pd.MultiIndex.from_arrays([[0.01, 0.02, 0.05, 0.1, 0.5],
                                 ['capacitance [pF]'] * 5])

cell_rise_Inv= pd.DataFrame([[0.00483211, 0.00511619, 0.00891821, 0.0449637, 0.205753], 
                             [0.00520049, 0.00561577, 0.010993, 0.0468998, 0.207461],
                             [0.00357213, 0.00429087, 0.0132186, 0.0536389, 0.21384],
                             [-0.0021868, -0.0011312, 0.0120546, 0.0647213, 0.224749],
                             [-0.0725403, -0.0700884, -0.0382486, 0.0899121, 0.313639]], 
                            index=idx,
                            columns=col)
cell_rise_Inv

您的代码与示例之间存在一个主要区别:该示例传递一个 numpy 数组作为输入,而不是嵌套列表。事实上,在您的列表周围添加 np.array(...) 效果很好:

cell_rise_Inv= pd.DataFrame(
        np.array([[0.00483211, 0.00511619, 0.00891821, 0.0449637, 0.205753], 
                  [0.00520049, 0.00561577, 0.010993, 0.0468998, 0.207461],
                  [0.00357213, 0.00429087, 0.0132186, 0.0536389, 0.21384],
                  [-0.0021868, -0.0011312, 0.0120546, 0.0647213, 0.224749],
                  [-0.0725403, -0.0700884, -0.0382486, 0.0899121, 0.313639]]), 
        index=[['transition [ns]'] * 5,
               [0.0005, 0.001, 0.01, 0.1, 0.5]],
        columns=[['capacitance [pF]'] * 5,
                 [0.01, 0.02, 0.05, 0.1, 0.5]])

我缩短了索引中重复的字符串并交换了索引级别的顺序,但这些都不是重大更改。

编辑 做了一点调查。如果你传入一个嵌套列表(没有 np.array 调用),调用将在没有 columns 的情况下工作,即使 columns 是一维列表。出于某种原因,两个元素的嵌套列表不会被解释为多重索引,除非输入是 ndarray.

我根据这个问题用 pandas 提交了 issue #14467