pandas知识点（数据结构）

1.Series

生成一维数组，左边索引，右边值：

In [3]: obj = Series([1,2,3,4,5])
In [4]: obj
Out[4]:
0    1
1    2
2    3
3    4
4    5
dtype: int64
In [5]: obj.values
Out[5]: array([1, 2, 3, 4, 5], dtype=int64)
In [6]: obj.index
Out[6]: RangeIndex(start=0, stop=5, step=1)

创建对各个数据点进行标记的索引：

In [7]: obj2 = Series([4,1,9,7], index=["a","c","e","ff"])
In [8]: obj2
Out[8]:
a     4
c     1
e     9
ff    7
dtype: int64
In [9]: obj2.index
Out[9]: Index(['a', 'c', 'e', 'ff'], dtype='object')

取一个值或一组值：

In [10]: obj2["c"]
Out[10]: 1
In [11]: obj2[["c","e"]]
Out[11]:
c    1
e    9
dtype: int64

数组运算，会显示索引：

In []: obj2[obj2>]
Out[]:
a
e
ff
dtype: int64

Series还可以看作有序的字典，很多字典操作可以使用：

In [13]: "c" in obj2
Out[13]: True

直接用字典创建Series：

In [14]: data = {"name":"liu","year":18,"sex":"man"}
In [15]: obj3 = Series(data)
In [16]: obj3
Out[16]:
name    liu
year     18
sex     man
dtype: object

用字典结合列表创建Series：

In [17]: list1 = ["name","year","mobile"]
In [18]: obj4 = Series(data,index=list1)
In [19]: obj4
Out[19]:
name      liu
year       18
mobile    NaN
dtype: object

PS：因为data字典中没有mobile所以值为NaN

 

检测数据是否缺失：

In [20]: pd.isnull(obj4)
Out[20]:
name      False
year      False
mobile     True
dtype: bool

In [21]: pd.notnull(obj4)
Out[21]:
name       True
year       True
mobile    False
dtype: bool

In [22]: obj4.isnull()
Out[22]:
name      False
year      False
mobile     True
dtype: bool

In [23]: obj4.notnull()
Out[23]:
name       True
year       True
mobile    False
dtype: bool

Series的name属性：

In [7]: obj4.name = "hahaha"
In [8]: obj4.index.name = "state"
In [9]: obj4
Out[9]:
state
name      liu
year       18
mobile    NaN
Name: hahaha, dtype: object

2.DataFrame

构建DataFrame

In [13]: data = {
"state":[1,1,2,1,1],
"year":[2000,2001,2002,2004,2005],
"pop":[1.5,1.7,3.6,2.4,2.9]
}
In [14]: frame = DataFrame(data)
In [15]: frame
Out[15]:
   state  year  pop
0      1  2000  1.5
1      1  2001  1.7
2      2  2002  3.6
3      1  2004  2.4
4      1  2005  2.9

设定行与列的名称，如果数据找不到则产生NA值：

In [18]: frame2 = DataFrame(
data,
columns=["year","state","pop","debt"],
index=["one","two","three","four","five"]
)
In [19]: frame2
Out[19]:
       year  state  pop debt
one    2000      1  1.5  NaN
two    2001      1  1.7  NaN
three  2002      2  3.6  NaN
four   2004      1  2.4  NaN
five   2005      1  2.9  NaN

将DataFrame的列获取成为Series：

In [7]: frame2.year
Out[7]:
one      2000
two      2001
three    2002
four     2004
five     2005
Name: year, dtype: int64

PS：返回的索引不变，且name属性被设置了

获取行：

In [11]: frame2.loc["three"]
Out[11]:
year     2002
state       2
pop       3.6
debt      NaN
Name: three, dtype: object

赋值列：

In [12]: frame2['debt'] = 16.5
In [13]: frame2
Out[13]:
       year  state  pop  debt
one    2000      1  1.5  16.5
two    2001      1  1.7  16.5
three  2002      2  3.6  16.5
four   2004      1  2.4  16.5
five   2005      1  2.9  16.5

如果赋值列表或数组，长度需要相等；如果赋值Series，则精确匹配索引

In [17]: val = Series([1.2,1.5,1.7], index=["two","four","five"])
In [18]: frame2['debt'] = val
In [19]: frame2
Out[19]:
       year  state  pop  debt
one    2000      1  1.5   NaN
two    2001      1  1.7   1.2
three  2002      2  3.6   NaN
four   2004      1  2.4   1.5
five   2005      1  2.9   1.7

如果列不存在，则创建：

In [21]: frame2["eastern"] = frame2.state == 1
In [22]: frame2
Out[22]:
       year  state  pop  debt  eastern
one    2000      1  1.5   NaN     True
two    2001      1  1.7   1.2     True
three  2002      2  3.6   NaN    False
four   2004      1  2.4   1.5     True
five   2005      1  2.9   1.7     True

对于嵌套字典，DataFrame会解释为外层为列，内层为行索引：

In [23]: dic = {"name":{"one":"liu","two":"rui"},"year":{"one":"","two":""}}
In [24]: frame3 = DataFrame(dic)
In [25]: frame3
Out[25]:
    name year
one  liu   23
two  rui   22

显示行，列名：

In [26]: frame3.index.name = "index"
In [27]: frame3.columns.name = "state"
In [28]: frame3
Out[28]:
state name year
index
one    liu   23
two    rui   22

返回二维ndarray形式的数据：

In [29]: frame3.values
Out[29]:
array([['liu', ''],
       ['rui', '']], dtype=object)

3.索引对象

In [30]: obj = Series(range(3),index=["a","b","c"])
In [31]: index = obj.index
In [32]: index
Out[32]: Index(['a', 'b', 'c'], dtype='object')

index对象不可修改的，使得index在多个数据结构中可以共享

In [35]: index = pd.Index(np.arange(3))
In [36]: obj2 = Series([1.5,0.5,2],index=index)
In [37]: obj2.index is index
Out[37]: True