pandas的數(shù)據(jù)結(jié)構(gòu)介紹
要使用pandas昼接,你首先就得熟悉它的兩個主要數(shù)據(jù)結(jié)構(gòu):Series和DataFrame炼邀。雖然它們并不能解決所有問題斤寇,但它們?yōu)榇蠖鄶?shù)應(yīng)用提供了一種可靠的、易于使用的基礎(chǔ)。
In [1]: from pandas import Series,DataFrame
In [2]: import pandas as pd
In [3]: import numpy as np
Series
Series是一種類似于一維數(shù)組的對象曲尸,它由一組數(shù)據(jù)(各種NumPy數(shù)據(jù)類型)以及一組與之相關(guān)的數(shù)據(jù)標(biāo)簽(即索引)組成。僅由一組數(shù)據(jù)即可產(chǎn)生最簡單的Series男翰。
In [4]: obj=Series([5,8,-6,2])
In [5]: obj
Out[5]:
0 5
1 8
2 -6
3 2
dtype: int64
Series的字符串表現(xiàn)形式為:索引在左邊另患,值在右邊。由于我們沒有為數(shù)據(jù)指定索引蛾绎,于是會自動創(chuàng)建一個0到N1(N為數(shù)據(jù)的長度)的整數(shù)型索引昆箕。你可以通過Series 的values和index屬性獲取其數(shù)組表示形式和索引對象鸦列。
In [6]: obj.values
Out[6]: array([ 5, 8, -6, 2], dtype=int64)
In [7]: obj.index
Out[7]: RangeIndex(start=0, stop=4, step=1)
希望所創(chuàng)建的Series帶有一個可以對各個數(shù)據(jù)點進(jìn)行標(biāo)記的索引
In [8]: obj2=Series([4,8,-6,3],index=['d','a','b','c'])
In [9]: obj2
Out[9]:
d 4
a 8
b -6
c 3
dtype: int64
與普通NumPy數(shù)組相比,你可以通過索引的方式選取Series中的單個或一組值
In [10]: obj2['b']
Out[10]: -6
In [11]: obj2['d']=9
In [12]: obj2[['c','a','d']]
Out[12]:
c 3
a 8
d 9
dtype: int64
NumPy數(shù)組運算(如根據(jù)布爾型數(shù)組進(jìn)行過濾鹏倘、標(biāo)量乘法薯嗤、應(yīng)用數(shù)學(xué)函數(shù)等)都會保留索引和值之間的鏈接
In [13]: obj2
Out[13]:
d 9
a 8
b -6
c 3
dtype: int64
In [14]: obj2[obj2>0]
Out[14]:
d 9
a 8
c 3
dtype: int64
In [15]: obj2*3
Out[15]:
d 27
a 24
b -18
c 9
dtype: int64
In [16]: np.exp(obj2)
Out[16]:
d 8103.083928
a 2980.957987
b 0.002479
c 20.085537
dtype: float64
可以將Series看成是一個定長的有序字典,因為它是索引值到數(shù)據(jù)值的一個映射纤泵。它可以用在許多原本需要字典參數(shù)的函數(shù)中骆姐。
In [17]: 'b' in obj2
Out[17]: True
In [18]: 'k' in obj2
Out[18]: False
如果數(shù)據(jù)被存放在一個Python字典中,也可以直接通過這個字典來創(chuàng)建Series
In [19]: sdata = {'Ohio': 35000, 'Texas': 71000, 'Oregon': 16000, 'Utah': 5000}
In [20]: obj3=Series(sdata)
In [21]: obj3
Out[21]:
Ohio 35000
Oregon 16000
Texas 71000
Utah 5000
dtype: int64
如果只傳入一個字典捏题,則結(jié)果Series中的索引就是原字典的鍵(有序排列)
In [22]: states = ['California', 'Ohio', 'Oregon', 'Texas']
In [23]: obj4 = Series(sdata, index=states)
In [24]: obj4
Out[24]:
California NaN
Ohio 35000.0
Oregon 16000.0
Texas 71000.0
dtype: float64
sdata中跟states索引相匹配的那3個值會被找出來并放到相應(yīng)的位置上玻褪,但由于"California"所對應(yīng)的sdata值找不到,所以其結(jié)果就為NaN(即“非數(shù)字”(not a number)公荧,在pandas中带射,它用于表示缺失或NA值)。我將使用缺失(missing)或NA表示缺失數(shù)據(jù)循狰。pandas的isnull和notnull函數(shù)可用于檢測缺失數(shù)據(jù)
In [25]: pd.isnull(obj4)
Out[25]:
California True
Ohio False
Oregon False
Texas False
dtype: bool
In [26]: pd.notnull(obj4)
Out[26]:
California False
Ohio True
Oregon True
Texas True
dtype: bool
In [27]: obj4.isnull()
Out[27]:
California True
Ohio False
Oregon False
Texas False
dtype: bool
Series最重要的一個功能是:它在算術(shù)運算中會自動對齊不同索引的數(shù)據(jù)窟社。
In [28]: obj3
Out[28]:
Ohio 35000
Oregon 16000
Texas 71000
Utah 5000
dtype: int64
In [29]: obj4
Out[29]:
California NaN
Ohio 35000.0
Oregon 16000.0
Texas 71000.0
dtype: float64
In [30]: obj3+obj4
Out[30]:
California NaN
Ohio 70000.0
Oregon 32000.0
Texas 142000.0
Utah NaN
dtype: float64
Series對象本身及其索引都有一個name屬性,該屬性跟pandas其他的關(guān)鍵功能關(guān)系非常密切
In [31]: obj4.name = 'population'
In [32]: obj4.index.name = 'state'
In [33]: obj4
Out[33]:
state
California NaN
Ohio 35000.0
Oregon 16000.0
Texas 71000.0
Name: population, dtype: float64
In [34]: obj
Out[34]:
0 5
1 8
2 -6
3 2
dtype: int64
Series的索引可以通過賦值的方式就地修改
In [35]: obj.index = ['Bob', 'Steve', 'Jeff', 'Ryan']
In [36]: obj
Out[36]:
Bob 5
Steve 8
Jeff -6
Ryan 2
dtype: int64
DataFrame
DataFrame是一個表格型的數(shù)據(jù)結(jié)構(gòu)绪钥,它含有一組有序的列桥爽,每列可以是不同的值類型(數(shù)值、字符串昧识、布爾值等)钠四。DataFrame既有行索引也有列索引,它可以被看做由Series組成的字典(共用同一個索引)跪楞。DataFrame中的數(shù)據(jù)是以一個或多個二維塊存放的(而不是列表缀去、字典或別的一維數(shù)據(jù)結(jié)構(gòu))。
注意: 雖然DataFrame是以二維結(jié)構(gòu)保存數(shù)據(jù)的甸祭,但你仍然可以輕松地將其表示為更高維度的數(shù)據(jù)缕碎。
構(gòu)建DataFrame的辦法有很多,最常用的一種是直接傳入一個由等長列表或NumPy數(shù)組組成的字典池户。
In [37]: data = {'state': ['Ohio', 'Ohio', 'Ohio', 'Nevada', 'Nevada'],
...: 'year': [2000, 2001, 2002, 2001, 2002],
...: 'pop': [1.5, 1.7, 3.6, 2.4, 2.9]}
In [38]: frame=DataFrame(data)
結(jié)果DataFrame會自動加上索引(跟Series一樣)咏雌,且全部列會被有序排列
In [39]: frame
Out[39]:
pop state year
0 1.5 Ohio 2000
1 1.7 Ohio 2001
2 3.6 Ohio 2002
3 2.4 Nevada 2001
4 2.9 Nevada 2002
如果指定了列序列,則DataFrame的列就會按照指定順序進(jìn)行排序
In [40]: DataFrame(data, columns=['year', 'state', 'pop'])
Out[40]:
year state pop
0 2000 Ohio 1.5
1 2001 Ohio 1.7
2 2002 Ohio 3.6
3 2001 Nevada 2.4
4 2002 Nevada 2.9
In [41]: frame2 = DataFrame(data, columns=['year', 'state', 'pop', 'debt'],
...: index=['one', 'two', 'three', 'four', 'five'])
跟Series一樣校焦,如果傳入的列在數(shù)據(jù)中找不到赊抖,就會產(chǎn)生NA值
In [42]: frame2
Out[42]:
year state pop debt
one 2000 Ohio 1.5 NaN
two 2001 Ohio 1.7 NaN
three 2002 Ohio 3.6 NaN
four 2001 Nevada 2.4 NaN
five 2002 Nevada 2.9 NaN
In [43]: frame2.columns
Out[43]: Index(['year', 'state', 'pop', 'debt'], dtype='object')
通過類似字典標(biāo)記的方式或?qū)傩缘姆绞剑梢詫ataFrame的列獲取為一個Series
In [44]: frame2['state']
Out[44]:
one Ohio
two Ohio
three Ohio
four Nevada
five Nevada
Name: state, dtype: object
In [45]: frame2.year
Out[45]:
one 2000
two 2001
three 2002
four 2001
five 2002
Name: year, dtype: int64
注意寨典,返回的Series擁有原DataFrame相同的索引氛雪,且其name屬性也已經(jīng)被相應(yīng)地設(shè)置好了。行也可以通過位置或名稱的方式進(jìn)行獲取耸成,比如用索引字段ix(被舍棄使用)报亩,可以使用loc
In [46]: frame2.ix['three']
C:\ProgramData\Anaconda3\lib\site-packages\ipykernel_launcher.py:1: DeprecationWarning:
.ix is deprecated. Please use
.loc for label based indexing or
.iloc for positional indexing
See the documentation here:
http://pandas.pydata.org/pandas-docs/stable/indexing.html#deprecate_ix
"""Entry point for launching an IPython kernel.
Out[46]:
year 2002
state Ohio
pop 3.6
debt NaN
Name: three, dtype: object
In [47]: frame2.loc['three']
Out[47]:
year 2002
state Ohio
pop 3.6
debt NaN
Name: three, dtype: object
列可以通過賦值的方式進(jìn)行修改浴鸿。例如,我們可以給那個空的"debt"列賦上一個標(biāo)量值或一組值
In [48]: frame2['debt'] = 18.5
In [49]: frame2
Out[49]:
year state pop debt
one 2000 Ohio 1.5 18.5
two 2001 Ohio 1.7 18.5
three 2002 Ohio 3.6 18.5
four 2001 Nevada 2.4 18.5
five 2002 Nevada 2.9 18.5
In [50]: frame2['debt'] = np.arange(5.)
In [51]: frame2
Out[51]:
year state pop debt
one 2000 Ohio 1.5 0.0
two 2001 Ohio 1.7 1.0
three 2002 Ohio 3.6 2.0
four 2001 Nevada 2.4 3.0
five 2002 Nevada 2.9 4.0
將列表或數(shù)組賦值給某個列時弦追,其長度必須跟DataFrame的長度相匹配岳链。如果賦值的是一個Series,就會精確匹配DataFrame的索引劲件,所有的空位都將被填上缺失值
In [52]: val = Series([-1.2, -1.5, -1.7], index=['two', 'four', 'five'])
In [53]: frame2['debt']=val
In [54]: frame2
Out[54]:
year state pop debt
one 2000 Ohio 1.5 NaN
two 2001 Ohio 1.7 -1.2
three 2002 Ohio 3.6 NaN
four 2001 Nevada 2.4 -1.5
five 2002 Nevada 2.9 -1.7
為不存在的列賦值會創(chuàng)建出一個新列宠页。關(guān)鍵字del用于刪除列
In [55]: frame2['eastern'] = frame2.state == 'Ohio'
In [56]: frame2
Out[56]:
year state pop debt eastern
one 2000 Ohio 1.5 NaN True
two 2001 Ohio 1.7 -1.2 True
three 2002 Ohio 3.6 NaN True
four 2001 Nevada 2.4 -1.5 False
five 2002 Nevada 2.9 -1.7 False
In [57]: del frame2['eastern']
In [58]: frame2.columns
Out[58]: Index(['year', 'state', 'pop', 'debt'], dtype='object')
警告: 通過索引方式返回的列只是相應(yīng)數(shù)據(jù)的視圖而已,并不是副本寇仓。因此究驴,對返回的Series所做的任何就地修改全都會反映到源DataFrame上磺送。通過Series的copy方法即可顯式地復(fù)制列。
另一種常見的數(shù)據(jù)形式是嵌套字典(也就是字典的字典),如果將它傳給DataFrame,它就會被解釋為:外層字典的鍵作為列地粪,內(nèi)層鍵則作為行索引家制。
In [64]: pop = {'Nevada': {2001: 2.4, 2002: 2.9},'Ohio': {2000: 1.5, 2001: 1.7, 2002: 3.6}}
In [65]: frame3=DataFrame(pop)
In [66]: frame3
Out[66]:
Nevada Ohio
2000 NaN 1.5
2001 2.4 1.7
2002 2.9 3.6
對該結(jié)果進(jìn)行轉(zhuǎn)置
In [67]: frame3.T
Out[67]:
2000 2001 2002
Nevada NaN 2.4 2.9
Ohio 1.5 1.7 3.6
內(nèi)層字典的鍵會被合并饼拍、排序以形成最終的索引渣磷。如果顯式指定了索引,則不會這樣
In [68]: DataFrame(pop, index=[2001, 2002, 2003])
Out[68]:
Nevada Ohio
2001 2.4 1.7
2002 2.9 3.6
2003 NaN NaN
由Series組成的字典差不多也是一樣的用法
In [69]: pdata = {'Ohio': frame3['Ohio'][:-1],'Nevada': frame3['Nevada'][:2]}
In [70]: DataFrame(pdata)
Out[70]:
Nevada Ohio
2000 NaN 1.5
2001 2.4 1.7
表5-1列出了DataFrame構(gòu)造函數(shù)所能接受的各種數(shù)據(jù)
如果設(shè)置了DataFrame的index和columns的name屬性罢猪,則這些信息也會被顯示出來
In [71]: frame3.index.name = 'year'; frame3.columns.name = 'state'
In [72]: frame3
Out[72]:
state Nevada Ohio
year
2000 NaN 1.5
2001 2.4 1.7
2002 2.9 3.6
跟Series一樣近她,values屬性也會以二維ndarray的形式返回DataFrame中的數(shù)據(jù)
In [73]: frame3.values
Out[73]:
array([[ nan, 1.5],
[ 2.4, 1.7],
[ 2.9, 3.6]])
如果DataFrame各列的數(shù)據(jù)類型不同,則值數(shù)組的數(shù)據(jù)類型就會選用能兼容所有列的數(shù)據(jù)類型
In [74]: frame2.values
Out[74]:
array([[2000, 'Ohio', 1.5, nan],
[2001, 'Ohio', 1.7, -1.2],
[2002, 'Ohio', 3.6, nan],
[2001, 'Nevada', 2.4, -1.5],
[2002, 'Nevada', 2.9, -1.7]], dtype=object)
索引對象
pandas的索引對象負(fù)責(zé)管理軸標(biāo)簽和其他元數(shù)據(jù)(比如軸名稱等)膳帕。構(gòu)建Series或DataFrame時粘捎,所用到的任何數(shù)組或其他序列的標(biāo)簽都會被轉(zhuǎn)換成一個Index:
In [75]: obj = Series(range(3), index=['a', 'b', 'c'])
In [76]: obj
Out[76]:
a 0
b 1
c 2
dtype: int32
In [77]: index=obj.index
In [78]: index
Out[78]: Index(['a', 'b', 'c'], dtype='object')
In [79]: index[1:]
Out[79]: Index(['b', 'c'], dtype='object')
Index對象是不可修改的(immutable),因此用戶不能對其進(jìn)行修改
In [80]: index[1]='d'
---------------------------------------------------------------------------
TypeError Traceback (most recent call last)
<ipython-input-80-d3f90986bdb1> in <module>()
----> 1 index[1]='d'
C:\ProgramData\Anaconda3\lib\site-packages\pandas\core\indexes\base.py in __setitem__(self, key, value)
1618
1619 def __setitem__(self, key, value):
-> 1620 raise TypeError("Index does not support mutable operations")
1621
1622 def __getitem__(self, key):
TypeError: Index does not support mutable operations
不可修改性非常重要危彩,因為這樣才能使Index對象在多個數(shù)據(jù)結(jié)構(gòu)之間安全共享
In [81]: index = pd.index(np.arange(3))
---------------------------------------------------------------------------
AttributeError Traceback (most recent call last)
<ipython-input-81-5eb0f1ff3872> in <module>()
----> 1 index = pd.index(np.arange(3))
AttributeError: module 'pandas' has no attribute 'index'
In [82]: index = pd.Index(np.arange(3))
In [83]: index
Out[83]: Int64Index([0, 1, 2], dtype='int64')
In [84]: obj2 = Series([1.5, -3.5, 0], index=index)
In [85]: obj2
Out[85]:
0 1.5
1 -3.5
2 0.0
dtype: float64
In [86]: obj2.index is index
Out[86]: True
表5-2列出了pandas庫中內(nèi)置的Index類
除了長得像數(shù)組攒磨,Index的功能也類似一個固定大小的集合
In [87]: frame3
Out[87]:
state Nevada Ohio
year
2000 NaN 1.5
2001 2.4 1.7
2002 2.9 3.6
In [88]: 'Ohio' in frame3.columns
Out[88]: True
In [89]: 2003 in frame3.index
Out[89]: False
每個索引都有一些方法和屬性,它們可用于設(shè)置邏輯并回答有關(guān)該索引所包含的數(shù)據(jù)的常見問題汤徽。表5-3列出了這些函數(shù)娩缰。
pandas數(shù)據(jù)結(jié)構(gòu)的按照書本上進(jìn)行練習(xí),通過練習(xí)操作谒府,可以更好的懂著這些基本的數(shù)據(jù)結(jié)構(gòu)拼坎。