pandas

最新推荐文章于 2022-09-09 15:18:21 发布

lls2012

最新推荐文章于 2022-09-09 15:18:21 发布

阅读量318

点赞数

CC 4.0 BY-SA版权

本文链接：https://blog.youkuaiyun.com/lls2012/article/details/78287501

In [3]:

import pandas as pd

import numpy as np

import matplotlib.pyplot as plt

In [7]:

s = pd.Series([1,2,3,np.nan,6,9])

Out[7]:

0 1.0

1 2.0

2 3.0

3 NaN

4 6.0

5 9.0

dtype: float64

In [ ]:

dates = pd.date_range('20170101', periods=6)

dates

Out[ ]:

DatetimeIndex(['2017-01-01', '2017-01-02', '2017-01-03','2017-01-04',

'2017-01-05','2017-01-06'],

dtype='datetime64[ns]', freq='D')

In [16]:

df = pd.DataFrame(np.random.randn(6, 4), index=dates, columns=list('ABCD'))

df

Out[16]:

A	B	C	D
2017-01-01	-0.107708	-1.024182	-0.678251	-2.229398
2017-01-02	0.922306	0.817418	0.152669	-0.009996
2017-01-03	-2.151562	-0.396791	-1.030035	-0.563381
2017-01-04	-0.656683	-0.877806	0.564612	0.846301
2017-01-05	-0.129575	-0.704444	-0.697426	-0.328053
2017-01-06	1.033851	0.355536	0.326386	-0.477157

In [19]:

df2 = pd.DataFrame({'A':1,

                    'B':pd.Timestamp('20130102'),

                    'C':pd.Series(1, index=list(range(4)),dtype='float32'),

                    'D':np.array([3]*4,dtype='int32'),

                    'E':pd.Categorical(['test', 'train', 'test', 'train']),

                    'F':'foo'})

df2

Out[19]:

A	B	C	D	E	F
0	1	2013-01-02	1.0	3	test	foo
1	1	2013-01-02	1.0	3	train	foo
2	1	2013-01-02	1.0	3	test	foo
3	1	2013-01-02	1.0	3	train	foo

In [20]:

df2.dtypes

Out[20]:

A int64

B datetime64[ns]

C float32

D int32

E category

F object

dtype: object

In [22]:

df.head()

Out[22]:

A	B	C	D
2017-01-01	-0.107708	-1.024182	-0.678251	-2.229398
2017-01-02	0.922306	0.817418	0.152669	-0.009996
2017-01-03	-2.151562	-0.396791	-1.030035	-0.563381
2017-01-04	-0.656683	-0.877806	0.564612	0.846301
2017-01-05	-0.129575	-0.704444	-0.697426	-0.328053

In [24]:

df.tail(3)

Out[24]:

A	B	C	D
2017-01-04	-0.656683	-0.877806	0.564612	0.846301
2017-01-05	-0.129575	-0.704444	-0.697426	-0.328053
2017-01-06	1.033851	0.355536	0.326386	-0.477157

In [25]:

df.index

Out[25]:

DatetimeIndex(['2017-01-01', '2017-01-02', '2017-01-03','2017-01-04',

'2017-01-05','2017-01-06'],

dtype='datetime64[ns]', freq='D')

In [27]:

df.columns

Out[27]:

Index(['A', 'B', 'C', 'D'], dtype='object')

In [28]:

df.values

Out[28]:

array([[-0.1077083 , -1.02418217, -0.67825098, -2.22939769],

[ 0.92230625, 0.81741758, 0.15266857, -0.00999563],

[-2.15156187,-0.39679118, -1.03003536, -0.56338102],

[-0.6566828 ,-0.8778064 , 0.56461162, 0.84630068],

[-0.12957523,-0.70444375, -0.69742558, -0.32805333],

[ 1.03385071, 0.35553587, 0.3263864 , -0.47715656]])

In [30]:

df.describe()

Out[30]:

A	B	C	D
count	6.000000	6.000000	6.000000	6.000000
mean	-0.181562	-0.305045	-0.227008	-0.460281
std	1.167095	0.736130	0.655260	1.006785
min	-2.151562	-1.024182	-1.030035	-2.229398
25%	-0.524906	-0.834466	-0.692632	-0.541825
50%	-0.118642	-0.550617	-0.262791	-0.402605
75%	0.664803	0.167454	0.282957	-0.089510
max	1.033851	0.817418	0.564612	0.846301

In [31]:

df.T

Out[31]:

2017-01-01 00:00:00	2017-01-02 00:00:00	2017-01-03 00:00:00	2017-01-04 00:00:00	2017-01-05 00:00:00	2017-01-06 00:00:00
A	-0.107708	0.922306	-2.151562	-0.656683	-0.129575	1.033851
B	-1.024182	0.817418	-0.396791	-0.877806	-0.704444	0.355536
C	-0.678251	0.152669	-1.030035	0.564612	-0.697426	0.326386
D	-2.229398	-0.009996	-0.563381	0.846301	-0.328053	-0.477157

In [46]:

df.sort_index(axis=1, ascending=False)

Out[46]:

D	C	B	A
2017-01-01	-2.229398	-0.678251	-1.024182	-0.107708
2017-01-02	-0.009996	0.152669	0.817418	0.922306
2017-01-03	-0.563381	-1.030035	-0.396791	-2.151562
2017-01-04	0.846301	0.564612	-0.877806	-0.656683
2017-01-05	-0.328053	-0.697426	-0.704444	-0.129575
2017-01-06	-0.477157	0.326386	0.355536	1.033851

In [64]:

df.sort_values(by='B')

Out[64]:

A	B	C	D
2017-01-01	-0.107708	-1.024182	-0.678251	-2.229398
2017-01-04	-0.656683	-0.877806	0.564612	0.846301
2017-01-05	-0.129575	-0.704444	-0.697426	-0.328053
2017-01-03	-2.151562	-0.396791	-1.030035	-0.563381
2017-01-06	1.033851	0.355536	0.326386	-0.477157
2017-01-02	0.922306	0.817418	0.152669	-0.009996

In [65]:

df['A']

Out[65]:

2017-01-01 -0.107708

2017-01-02 0.922306

2017-01-03 -2.151562

2017-01-04 -0.656683

2017-01-05 -0.129575

2017-01-06 1.033851

Freq: D, Name: A, dtype: float64

In [66]:

df[0:3]

Out[66]:

A	B	C	D
2017-01-01	-0.107708	-1.024182	-0.678251	-2.229398
2017-01-02	0.922306	0.817418	0.152669	-0.009996
2017-01-03	-2.151562	-0.396791	-1.030035	-0.563381

In [69]:

df.loc[dates[0]]

Out[69]:

A -0.107708

B -1.024182

C -0.678251

D -2.229398

Name: 2017-01-01 00:00:00, dtype: float64

In [70]:

df.loc[:, 'A':'B']

Out[70]:

A	B
2017-01-01	-0.107708	-1.024182
2017-01-02	0.922306	0.817418
2017-01-03	-2.151562	-0.396791
2017-01-04	-0.656683	-0.877806
2017-01-05	-0.129575	-0.704444
2017-01-06	1.033851	0.355536

In [71]:

df.loc['20170101':'20170103', 'A':'B']

Out[71]:

A	B
2017-01-01	-0.107708	-1.024182
2017-01-02	0.922306	0.817418
2017-01-03	-2.151562	-0.396791

In [72]:

df.loc['20170104', 'A':'B']

Out[72]:

A -0.656683

B -0.877806

Name: 2017-01-04 00:00:00, dtype: float64

In [73]:

df.loc['20170105','C']

Out[73]:

-0.69742558177888991

In [76]:

df.iloc[3]

Out[76]:

A -0.656683

B -0.877806

C 0.564612

D 0.846301

Name: 2017-01-04 00:00:00, dtype: float64

In [77]:

df.iloc[3:5, 0:2]

Out[77]:

A	B
2017-01-04	-0.656683	-0.877806
2017-01-05	-0.129575	-0.704444

In [79]:

df.iloc[[1,3,5],[0,3]]

Out[79]:

A	D
2017-01-02	0.922306	-0.009996
2017-01-04	-0.656683	0.846301
2017-01-06	1.033851	-0.477157

In [80]:

df.iloc[:,1:3]

Out[80]:

B	C
2017-01-01	-1.024182	-0.678251
2017-01-02	0.817418	0.152669
2017-01-03	-0.396791	-1.030035
2017-01-04	-0.877806	0.564612
2017-01-05	-0.704444	-0.697426
2017-01-06	0.355536	0.326386

In [81]:

df.iloc[1,1]

Out[81]:

0.81741758295028633

In [82]:

df.iat[1,1]

Out[82]:

0.81741758295028633

In [83]:

df[df.A>0]

Out[83]:

A	B	C	D
2017-01-02	0.922306	0.817418	0.152669	-0.009996
2017-01-06	1.033851	0.355536	0.326386	-0.477157

In [84]:

df[df>0]

Out[84]:

A	B	C	D
2017-01-01	NaN	NaN	NaN	NaN
2017-01-02	0.922306	0.817418	0.152669	NaN
2017-01-03	NaN	NaN	NaN	NaN
2017-01-04	NaN	NaN	0.564612	0.846301
2017-01-05	NaN	NaN	NaN	NaN
2017-01-06	1.033851	0.355536	0.326386	NaN

In [86]:

df3=df.copy()

df3['E']=['one','one', 'two','three','four','five']

df3

Out[86]:

A	B	C	D	E
2017-01-01	-0.107708	-1.024182	-0.678251	-2.229398	one
2017-01-02	0.922306	0.817418	0.152669	-0.009996	one
2017-01-03	-2.151562	-0.396791	-1.030035	-0.563381	two
2017-01-04	-0.656683	-0.877806	0.564612	0.846301	three
2017-01-05	-0.129575	-0.704444	-0.697426	-0.328053	four
2017-01-06	1.033851	0.355536	0.326386	-0.477157	five

In [88]:

df3[df3['E'].isin(['two', 'four'])]

Out[88]:

A	B	C	D	E
2017-01-03	-2.151562	-0.396791	-1.030035	-0.563381	two
2017-01-05	-0.129575	-0.704444	-0.697426	-0.328053	four

设置 #### 1、设置一个新的列:

In [93]:

s1 = pd.Series([1,2,3,4,5,6],index=pd.date_range('20170102',periods=6))

Out[93]:

2017-01-02 1

2017-01-03 2

2017-01-04 3

2017-01-05 4

2017-01-06 5

2017-01-07 6

Freq: D, dtype: int64

2、通过标签设置新的值:¶

In [94]:

df['F'] = s1

Out[94]:

A	B	C	D	F
2017-01-01	-0.107708	-1.024182	-0.678251	-2.229398	NaN
2017-01-02	0.922306	0.817418	0.152669	-0.009996	1.0
2017-01-03	-2.151562	-0.396791	-1.030035	-0.563381	2.0
2017-01-04	-0.656683	-0.877806	0.564612	0.846301	3.0
2017-01-05	-0.129575	-0.704444	-0.697426	-0.328053	4.0
2017-01-06	1.033851	0.355536	0.326386	-0.477157	5.0

In [95]:

df.at[dates[0], 'A'] = 0

Out[95]:

A	B	C	D	F
2017-01-01	0.000000	-1.024182	-0.678251	-2.229398	NaN
2017-01-02	0.922306	0.817418	0.152669	-0.009996	1.0
2017-01-03	-2.151562	-0.396791	-1.030035	-0.563381	2.0
2017-01-04	-0.656683	-0.877806	0.564612	0.846301	3.0
2017-01-05	-0.129575	-0.704444	-0.697426	-0.328053	4.0
2017-01-06	1.033851	0.355536	0.326386	-0.477157	5.0

In [97]:

df.iat[0,1] = 0

Out[97]:

A	B	C	D	F
2017-01-01	0.000000	0.000000	-0.678251	-2.229398	NaN
2017-01-02	0.922306	0.817418	0.152669	-0.009996	1.0
2017-01-03	-2.151562	-0.396791	-1.030035	-0.563381	2.0
2017-01-04	-0.656683	-0.877806	0.564612	0.846301	3.0
2017-01-05	-0.129575	-0.704444	-0.697426	-0.328053	4.0
2017-01-06	1.033851	0.355536	0.326386	-0.477157	5.0

4、通过numpy数组设置一组新值：¶

In [99]:

df.loc[:,'D'] = np.array([5]*len(df))

Out[99]:

A	B	C	D	F
2017-01-01	0.000000	0.000000	-0.678251	5	NaN
2017-01-02	0.922306	0.817418	0.152669	5	1.0
2017-01-03	-2.151562	-0.396791	-1.030035	5	2.0
2017-01-04	-0.656683	-0.877806	0.564612	5	3.0
2017-01-05	-0.129575	-0.704444	-0.697426	5	4.0
2017-01-06	1.033851	0.355536	0.326386	5	5.0

5、通过where操作来设置新的值：¶

In [101]:

df4 = df.copy()

df4[df4 >0] = -df4

df4

Out[101]:

A	B	C	D	F
2017-01-01	0.000000	0.000000	-0.678251	-5	NaN
2017-01-02	-0.922306	-0.817418	-0.152669	-5	-1.0
2017-01-03	-2.151562	-0.396791	-1.030035	-5	-2.0
2017-01-04	-0.656683	-0.877806	-0.564612	-5	-3.0
2017-01-05	-0.129575	-0.704444	-0.697426	-5	-4.0
2017-01-06	-1.033851	-0.355536	-0.326386	-5	-5.0

四、缺失值处理¶

在pandas中，使用np.nan来代替缺失值，这些值将默认不会包含在计算中，详细请参阅：MissingData Section

1、reindex()方法可以对指定轴上的索引进行改变/增加/删除操作，这将返回原始数据的一个拷贝：¶

In [105]:

df1 = df.reindex(index=dates[0:4], columns=list(df.columns) +['E'])

df1

Out[105]:

A	B	C	D	F	E
2017-01-01	0.000000	0.000000	-0.678251	5	NaN	NaN
2017-01-02	0.922306	0.817418	0.152669	5	1.0	NaN
2017-01-03	-2.151562	-0.396791	-1.030035	5	2.0	NaN
2017-01-04	-0.656683	-0.877806	0.564612	5	3.0	NaN

In [106]:

df1.loc[dates[0]:dates[1], 'E'] = 1

df1

Out[106]:

A	B	C	D	F	E
2017-01-01	0.000000	0.000000	-0.678251	5	NaN	1.0
2017-01-02	0.922306	0.817418	0.152669	5	1.0	1.0
2017-01-03	-2.151562	-0.396791	-1.030035	5	2.0	NaN
2017-01-04	-0.656683	-0.877806	0.564612	5	3.0	NaN

2、去掉包含缺失值的行：¶

In [107]:

df1.dropna(how='any')

Out[107]:

A	B	C	D	F	E
2017-01-02	0.922306	0.817418	0.152669	5	1.0	1.0

3、对缺失值进行填充：¶

In [108]:

df1.fillna(value=5)

Out[108]:

A	B	C	D	F	E
2017-01-01	0.000000	0.000000	-0.678251	5	5.0	1.0
2017-01-02	0.922306	0.817418	0.152669	5	1.0	1.0
2017-01-03	-2.151562	-0.396791	-1.030035	5	2.0	5.0
2017-01-04	-0.656683	-0.877806	0.564612	5	3.0	5.0

4、对数据进行布尔填充：¶

In [109]:

pd.isnull(df)

Out[109]:

A	B	C	D	F
2017-01-01	False	False	False	False	True
2017-01-02	False	False	False	False	False
2017-01-03	False	False	False	False	False
2017-01-04	False	False	False	False	False
2017-01-05	False	False	False	False	False
2017-01-06	False	False	False	False	False

五、相关操作¶

详情请参阅BasicSection OnBinary Ops

统计（相关操作通常情况不包括缺失值）

1、执行描述性统计：¶

In [112]:

df.mean()

Out[112]:

A -0.163610

B -0.134348

C -0.227008

D 5.000000

F 3.000000

dtype: float64

2、在其它轴上进行相同操作：¶

In [113]:

df.mean(1)

Out[113]:

2017-01-01 1.080437

2017-01-02 1.578478

2017-01-03 0.684322

2017-01-04 1.406024

2017-01-05 1.493711

2017-01-06 2.343155

Freq: D, dtype: float64

3、对于拥有不同维度，需要对齐的对象进行操作。Pandas会自动的沿着指定的维度进行广播：¶

In [120]:

s3 =pd.Series([1,3,4,np.nan,6,8], index=dates).shift(1)

Out[120]:

2017-01-01 NaN

2017-01-02 1.0

2017-01-03 3.0

2017-01-04 4.0

2017-01-05 NaN

2017-01-06 6.0

Freq: D, dtype: float64

In [123]:

df.sub(s3, axis='index')

Out[123]:

A	B	C	D	F
2017-01-01	NaN	NaN	NaN	NaN	NaN
2017-01-02	-0.077694	-0.182582	-0.847331	4.0	0.0
2017-01-03	-5.151562	-3.396791	-4.030035	2.0	-1.0
2017-01-04	-4.656683	-4.877806	-3.435388	1.0	-1.0
2017-01-05	NaN	NaN	NaN	NaN	NaN
2017-01-06	-4.966149	-5.644464	-5.673614	-1.0	-1.0

In [122]:

Out[122]:

A	B	C	D	F
2017-01-01	0.000000	0.000000	-0.678251	5	NaN
2017-01-02	0.922306	0.817418	0.152669	5	1.0
2017-01-03	-2.151562	-0.396791	-1.030035	5	2.0
2017-01-04	-0.656683	-0.877806	0.564612	5	3.0
2017-01-05	-0.129575	-0.704444	-0.697426	5	4.0
2017-01-06	1.033851	0.355536	0.326386	5	5.0

In [126]:

df.apply(np.cumsum)

Out[126]:

A	B	C	D	F
2017-01-01	0.000000	0.000000	-0.678251	5	NaN
2017-01-02	0.922306	0.817418	-0.525582	10	1.0
2017-01-03	-1.229256	0.420626	-1.555618	15	3.0
2017-01-04	-1.885938	-0.457180	-0.991006	20	6.0
2017-01-05	-2.015514	-1.161624	-1.688432	25	10.0
2017-01-06	-0.981663	-0.806088	-1.362045	30	15.0

In [127]:

df.apply(lambda x:x.max() - x.min())

Out[127]:

A 3.185413

B 1.695224

C 1.594647

D 0.000000

F 4.000000

dtype: float64

In [128]:

s = pd.Series(np.random.randint(0, 7, size=10))

Out[128]:

0 6

1 1

2 3

3 6

4 5

5 3

6 3

7 3

8 5

9 4

dtype: int32

In [129]:

s.value_counts()

Out[129]:

3 4

6 2

5 2

4 1

1 1

dtype: int64

In [130]:

s = pd.Series(['A', 'B', 'C', 'Aaba', 'Baca', np.nan, 'CABA','dog', 'cat'])

s.str.lower()

Out[130]:

0 a

1 b

2 c

3 aaba

4 baca

5 NaN

6 caba

7 dog

8 cat

dtype: object

In [131]:

df = pd.DataFrame(np.random.randn(10, 4))

Out[131]:

0	1	2	3
0	0.112665	0.048688	0.202798	0.025471
1	-1.390846	-0.538326	-0.078290	-0.995754
2	2.089888	0.442472	0.038945	-1.785647
3	-0.063596	-0.104295	1.330084	0.673432
4	-1.090966	-1.106314	1.203585	0.752954
5	0.650855	-0.551656	-0.476231	0.874108
6	0.515420	-0.099656	-1.028804	0.163295
7	-1.587624	-1.178458	0.869725	0.553387
8	-1.547900	-0.387638	0.458020	-0.131574
9	0.696191	-0.826334	0.482097	-1.577040

In [135]:

pieces = [df[:3], df[3:7], df[7:]]

pieces

Out[135]:

[ 0 1 2 3

0 0.112665 0.048688 0.202798 0.025471

1 -1.390846 -0.538326-0.078290 -0.995754

2 2.089888 0.442472 0.038945 -1.785647,

0 1 2 3

3 -0.063596 -0.104295 1.330084 0.673432

4 -1.090966 -1.106314 1.203585 0.752954

5 0.650855 -0.551656 -0.476231 0.874108

6 0.515420 -0.099656 -1.028804 0.163295,

0 1 2 3

7 -1.587624 -1.178458 0.869725 0.553387

8 -1.547900 -0.387638 0.458020 -0.131574

9 0.696191 -0.826334 0.482097 -1.577040]

In [136]:

pd.concat(pieces)

Out[136]:

0	1	2	3
0	0.112665	0.048688	0.202798	0.025471
1	-1.390846	-0.538326	-0.078290	-0.995754
2	2.089888	0.442472	0.038945	-1.785647
3	-0.063596	-0.104295	1.330084	0.673432
4	-1.090966	-1.106314	1.203585	0.752954
5	0.650855	-0.551656	-0.476231	0.874108
6	0.515420	-0.099656	-1.028804	0.163295
7	-1.587624	-1.178458	0.869725	0.553387
8	-1.547900	-0.387638	0.458020	-0.131574
9	0.696191	-0.826334	0.482097	-1.577040

In [137]:

left = pd.DataFrame({'key':['foo', 'foo'], 'lval':[1,2]})

right = pd.DataFrame({'key':['foo', 'foo'], 'rval':[4,5]})

left

Out[137]:

key	lval
0	foo	1
1	foo	2

In [138]:

right

Out[138]:

key	rval
0	foo	4
1	foo	5

In [140]:

pd.merge(left, right, on='key')

Out[140]:

key	lval	rval
0	foo	1	4
1	foo	1	5
2	foo	2	4
3	foo	2	5

In [141]:

df = pd.DataFrame(np.random.randn(8,4),columns=['A','B','C','D'])

Out[141]:

A	B	C	D
0	-0.277390	-0.004491	0.328250	-0.752631
1	1.246361	-0.560859	0.140553	-2.784473
2	-0.777135	-1.778617	1.686014	-0.227608
3	-0.016098	0.530940	-1.249833	1.977533
4	-1.018847	0.540397	-0.241101	-1.287702
5	1.836941	1.422687	0.376302	-0.804313
6	0.279718	0.694741	0.363781	1.292481
7	0.575892	-2.357896	0.781150	0.789141

In [142]:

s = df.loc[3]

df.append(s, ignore_index=True)

Out[142]:

A	B	C	D
0	-0.277390	-0.004491	0.328250	-0.752631
1	1.246361	-0.560859	0.140553	-2.784473
2	-0.777135	-1.778617	1.686014	-0.227608
3	-0.016098	0.530940	-1.249833	1.977533
4	-1.018847	0.540397	-0.241101	-1.287702
5	1.836941	1.422687	0.376302	-0.804313
6	0.279718	0.694741	0.363781	1.292481
7	0.575892	-2.357896	0.781150	0.789141
8	-0.016098	0.530940	-1.249833	1.977533

In [143]:

df = pd.DataFrame({'A':['foo', 'bar', 'foo', 'bar',

'foo', 'bar', 'foo', 'bar'],

'B':['one', 'one', 'two', 'three',

'two', 'two', 'one', 'three'],

'C':np.random.randn(8),

'D':np.random.randn(8)})

Out[143]:

A	B	C	D
0	foo	one	-0.200720	-2.172299
1	bar	one	0.701414	1.142280
2	foo	two	-0.384221	-0.842057
3	bar	three	-0.883506	-0.683085
4	foo	two	1.438267	0.485135
5	bar	two	0.252307	1.145261
6	foo	one	0.572104	0.869513
7	bar	three	0.760147	0.056735

In [146]:

df.groupby('A').sum()

Out[146]:

C	D
A
bar	0.830361	1.661191
foo	1.425430	-1.659709

In [147]:

df.groupby(['A', 'B']).sum()

Out[147]:

C	D
A	B
bar	one	0.701414	1.142280
	three	-0.123359	-0.626350
	two	0.252307	1.145261
foo	one	0.371384	-1.302786
foo	two	1.054046	-0.356923

In [158]:

tuples = list(zip(*[['bar', 'bar', 'baz', 'baz',

'foo', 'foo', 'qux', 'qux'],

['one', 'two', 'one', 'two',

'one', 'two', 'one', 'two']]))

tuples

Out[158]:

[('bar', 'one'),

('bar', 'two'),

('baz', 'one'),

('baz', 'two'),

('foo', 'one'),

('foo', 'two'),

('qux', 'one'),

('qux', 'two')]

In [159]:

index = pd.MultiIndex.from_tuples(tuples, names=['first','second'])

index

Out[159]:

MultiIndex(levels=[['bar', 'baz', 'foo', 'qux'], ['one', 'two']],

labels=[[0, 0, 1,1, 2, 2, 3, 3], [0, 1, 0, 1, 0, 1, 0, 1]],

names=['first','second'])

In [163]:

df = pd.DataFrame(np.random.randn(8, 2), index=index,columns=['A', 'B'])

Out[163]:

A	B
first	second
bar	one	2.496134	-0.893512
bar	two	2.286660	-0.248709
baz	one	0.147845	0.985195
baz	two	-0.529439	-0.095119
foo	one	0.618096	-0.364225
foo	two	-0.090616	0.032860
qux	one	-0.912772	-0.715831
qux	two	-1.164991	-1.050507

In [164]:

stacked = df.stack()

stacked

Out[164]:

first second

bar one A 2.496134

B -0.893512

two A 2.286660

B -0.248709

baz one A 0.147845

B 0.985195

two A -0.529439

B -0.095119

foo one A 0.618096

B -0.364225

two A -0.090616

B 0.032860

qux one A -0.912772

B -0.715831

two A -1.164991

B -1.050507

dtype: float64

In [166]:

stacked.unstack()

Out[166]:

A	B
first	second
bar	one	2.496134	-0.893512
bar	two	2.286660	-0.248709
baz	one	0.147845	0.985195
baz	two	-0.529439	-0.095119
foo	one	0.618096	-0.364225
foo	two	-0.090616	0.032860
qux	one	-0.912772	-0.715831
qux	two	-1.164991	-1.050507

In [167]:

stacked.unstack(1)

Out[167]:

second	one	two
first
bar	A	2.496134	2.286660
bar	B	-0.893512	-0.248709
baz	A	0.147845	-0.529439
baz	B	0.985195	-0.095119
foo	A	0.618096	-0.090616
foo	B	-0.364225	0.032860
qux	A	-0.912772	-1.164991
qux	B	-0.715831	-1.050507

In [168]:

stacked.unstack(0)

Out[168]:

first	bar	baz	foo	qux
second
one	A	2.496134	0.147845	0.618096	-0.912772
one	B	-0.893512	0.985195	-0.364225	-0.715831
two	A	2.286660	-0.529439	-0.090616	-1.164991
two	B	-0.248709	-0.095119	0.032860	-1.050507

In [ ]: