索引和旋转

object.ix[]支持整数和标签混合使用，但是pandas社区目前对这种语法持反对态度，转而使用严格的[]、loc、iloc语法

首先生成数据对象

>>> dates = pd.date_range('2020-9-14', periods=8)
>>> df = pd.DataFrame(np.random.randn(8,4), index=dates, columns=['a','b','c','d'])
>>> dfa	b	c	d
2020-09-14	-0.100720	1.637659	0.983527	0.844038
2020-09-15	-1.103113	0.791526	0.459550	-0.045499
2020-09-16	0.926683	-1.379218	-0.507922	-1.030637
2020-09-17	0.647245	-0.960574	0.572754	-1.431501
2020-09-18	1.281217	1.079646	-1.268456	0.406251
2020-09-19	-0.873806	0.415606	0.182284	-2.042309
2020-09-20	-0.584982	-0.741095	-0.342637	-1.595636
2020-09-21	-1.757857	1.389785	1.102973	-1.790435

• []语法：

>>> s = df['a']
>>> s
2020-09-14    1.194543
2020-09-15   -1.760886
2020-09-16   -1.280667
2020-09-17    0.202355
2020-09-18   -1.771945
2020-09-19    0.645342
2020-09-20   -1.215266
2020-09-21    0.039558
Freq: D, Name: a, dtype: float64>>> s[dates[5]] #等价于s[5]
0.6453420970434852>>> df[['b', 'a']] # 可以传一列参数b	a
2020-09-14	0.256071	1.194543
2020-09-15	-0.740801	-1.760886
2020-09-16	0.137337	-1.280667
2020-09-17	1.006071	0.202355
2020-09-18	0.196233	-1.771945
2020-09-19	-1.108175	0.645342
2020-09-20	-0.277499	-1.215266
2020-09-21	-0.076766	0.039558

• loc语法：严格基于标签
  • 一个标签：如’a’or1（1被解释成索引的标签）
  • 一列标签：如[‘a’,b’,‘c’]
  • 标签的分片，开始和结束都要包含
  • 单参数的可调用函数对象

使用数字会报错

>>> df.loc[1]
TypeError                                 Traceback (most recent call last)
<ipython-input-113-1f09d9ff6611> in <module>
----> 1 df.loc[1]~\scoop\apps\anaconda3\2020.07\lib\site-packages\pandas\core\indexing.py in __getitem__(self, key)1766 1767             maybe_callable = com.apply_if_callable(key, self.obj)
-> 1768             return self._getitem_axis(maybe_callable, axis=axis)1769 1770     def _is_scalar_access(self, key: Tuple):~\scoop\apps\anaconda3\2020.07\lib\site-packages\pandas\core\indexing.py in _getitem_axis(self, key, axis)1962 1963         # fall thru to straight lookup
-> 1964         self._validate_key(key, axis)1965         return self._get_label(key, axis=axis)1966 ~\scoop\apps\anaconda3\2020.07\lib\site-packages\pandas\core\indexing.py in _validate_key(self, key, axis)1829 1830         if not is_list_like_indexer(key):
-> 1831             self._convert_scalar_indexer(key, axis)1832 1833     def _is_scalar_access(self, key: Tuple) -> bool:~\scoop\apps\anaconda3\2020.07\lib\site-packages\pandas\core\indexing.py in _convert_scalar_indexer(self, key, axis)739         ax = self.obj._get_axis(min(axis, self.ndim - 1))740         # a scalar
--> 741         return ax._convert_scalar_indexer(key, kind=self.name)742 743     def _convert_slice_indexer(self, key: slice, axis: int):~\scoop\apps\anaconda3\2020.07\lib\site-packages\pandas\core\indexes\datetimelike.py in _convert_scalar_indexer(self, key, kind)384             is_flt = is_float(key)385             if kind in ["loc"] and (is_int or is_flt):
--> 386                 self._invalid_indexer("index", key)387             elif kind in ["ix", "getitem"] and is_flt:388                 self._invalid_indexer("index", key)~\scoop\apps\anaconda3\2020.07\lib\site-packages\pandas\core\indexes\base.py in _invalid_indexer(self, form, key)3074         Consistent invalid indexer message.3075         """
-> 3076         raise TypeError(3077             f"cannot do {form} indexing on {type(self)} with these "3078             f"indexers [{key}] of {type(key)}"TypeError: cannot do index indexing on <class 'pandas.core.indexes.datetimes.DatetimeIndex'> with these indexers [1] of <class 'int'>

• iloc是基于位置的，可以使用bool数组进行筛选。可接收：
  • 一个整数：5
  • 一个整数序列：[1,2,3]
  • 一个分片对象：1:7
  • 一个布尔数组

算数运算和对齐

pandas的重要特点是可以对两个索引不相同的对象进行运算。对于独有的，在结果里表示NaN，对于公共的给出运算结果。

>>> a = pd.DataFrame(np.arange(9).reshape(3,3), columns=tuple('abc'), index=[1,2,3])
>>> aa	b	c
1	0	1	2
2	3	4	5
3	6	7	8>>> b = pd.DataFrame(np.arange(9).reshape(3,3), columns=tuple('adc'), index=[1,2,3])
>>> b
a	d	c
1	0	1	2
2	3	4	5
3	6	7	8>>> a+ba	b	c	d
1	0	NaN	4	NaN
2	6	NaN	10	NaN
3	12	NaN	16	NaN

数据框和序列之间也可以进行算术运算，这时会用到数组的扩充转换机制。默认情况下序列的索引值和数据框的索引列表相匹配，并自动扩充到各行。

>>> aa	b	c
1	0	1	2
2	3	4	5
3	6	7	8>>> a.loc[1]
a    0
b    1
c    2
Name: 1, dtype: int32>>> a-a.loc[1]a	b	c
1	0	0	0
2	3	3	3
3	6	6	6