数据科学包二：pandas基础及进阶（一）

先马
官方文档：https://pandas.pydata.org/pandas-docs/stable/getting_started/index.html

dataframework的创建

import pandas as pd
import numpy as np
a=pd.Series([1,3,5])#创建一个一维数组，注意Series大写
#二维数组DataFram的创建
date=pd.date_range(start='2012',end='2016',periods=5)
col=['A','B','C','D']
sht=pd.DataFrame(np.random.randn(4,5),index=col,columns=date)
sht
Out[14]: 
   2012-01-01  ...  2016-01-01
A   -1.208135  ...    1.209262
B    0.098231  ...    0.145908
C    1.247936  ...    1.124116
D    0.263908  ...   -1.422481
[4 rows x 5 columns]


#二维表也可以用字典创建,key值表示colums
a={'TALL':[180,178,188],
   'GENDER':'MALE',
   'GRADE':np.array([90,95,96])}
b=pd.DataFrame(a,index=['tom','mike','bob'])

dataframework的简单操作

b.describe()#描述性统计
Out[16]: 
             TALL      GRADE
count    3.000000   3.000000
mean   182.000000  93.666667
std      5.291503   3.214550
min    178.000000  90.000000
25%    179.000000  92.500000
50%    180.000000  95.000000

b.T#转置
Out[17]: 
         tom  mike   bob
TALL     180   178   188
GENDER  MALE  MALE  MALE
GRADE     90    95    96

b.values#显示数值
Out[22]: 
array([[180, 'MALE', 90],
       [178, 'MALE', 95],
       [188, 'MALE', 96]], dtype=object)
       
b.index#显示行标签
Out[23]: Index(['tom', 'mike', 'bob'], dtype='object')

b.TALL#通过列标签查看此列数据
Out[27]: 
tom     180
mike    178
bob     188
Name: TALL, dtype: int64

数据查看

sht
Out[29]: 
   2012-01-01  ...  2016-01-01
A   -1.208135  ...    1.209262
B    0.098231  ...    0.145908
C    1.247936  ...    1.124116
D    0.263908  ...   -1.422481
[4 rows x 5 columns]
sht.head(2)#前两行
Out[30]: 
   2012-01-01  ...  2016-01-01
A   -1.208135  ...    1.209262
B    0.098231  ...    0.145908
[2 rows x 5 columns]
sht.tail(1)#后一行
Out[31]: 
   2012-01-01  ...  2016-01-01
D    0.263908  ...   -1.422481
[1 rows x 5 columns]

排序

# 首先创建一个sheet
sht
Out[60]: 
        2012-12-31  2013-12-31  2014-12-31  2015-12-31  2016-12-31
BOB      -0.804521    0.163748    0.321372    0.321230   -1.568735
ARON     -1.782169    0.471212    1.211516   -0.002611    1.971601
CARTER    0.490189    2.341276   -1.872003    0.422649    1.700321
DAVID     0.082172    1.387305   -0.142821   -0.616505    1.235456


sht.sort_index(axis=1,ascending=False)#列标签排序，降序
Out[61]: 
        2016-12-31  2015-12-31  2014-12-31  2013-12-31  2012-12-31
BOB      -1.568735    0.321230    0.321372    0.163748   -0.804521
ARON      1.971601   -0.002611    1.211516    0.471212   -1.782169
CARTER    1.700321    0.422649   -1.872003    2.341276    0.490189
DAVID     1.235456   -0.616505   -0.142821    1.387305    0.082172


sht.sort_index(axis=0)#行标签排序，升序
Out[62]: 
        2012-12-31  2013-12-31  2014-12-31  2015-12-31  2016-12-31
ARON     -1.782169    0.471212    1.211516   -0.002611    1.971601
BOB      -0.804521    0.163748    0.321372    0.321230   -1.568735
CARTER    0.490189    2.341276   -1.872003    0.422649    1.700321
DAVID     0.082172    1.387305   -0.142821   -0.616505    1.235456

sht.sort_values(by='2012-12-31')#按某列的值进行排序
Out[65]: 
        2012-12-31  2013-12-31  2014-12-31  2015-12-31  2016-12-31
ARON     -1.782169    0.471212    1.211516   -0.002611    1.971601
BOB      -0.804521    0.163748    0.321372    0.321230   -1.568735
DAVID     0.082172    1.387305   -0.142821   -0.616505    1.235456
CARTER    0.490189    2.341276   -1.872003    0.422649    1.700321

sht.sort_values(by='ARON',axis=1)#按某行的值进行排序
Out[68]: 
        2012-12-31  2015-12-31  2013-12-31  2014-12-31  2016-12-31
BOB      -0.804521    0.321230    0.163748    0.321372   -1.568735
ARON     -1.782169   -0.002611    0.471212    1.211516    1.971601
CARTER    0.490189    0.422649    2.341276   -1.872003    1.700321
DAVID     0.082172   -0.616505    1.387305   -0.142821    1.235456

loc/iloc函数（dataframework的灵活截取）

按照行标签/序数索引

sht.loc['BOB':'ARON']#标签索引
Out[72]: 
      2012-12-31  2013-12-31  2014-12-31  2015-12-31  2016-12-31
BOB    -0.804521    0.163748    0.321372    0.321230   -1.568735
ARON   -1.782169    0.471212    1.211516   -0.002611    1.971601

sht.iloc[2:3]#序数索引
Out[69]: 
        2012-12-31  2013-12-31  2014-12-31  2015-12-31  2016-12-31
CARTER    0.490189    2.341276   -1.872003    0.422649    1.700321


sht.iloc[0:2,1:3]#前两行及二三列
Out[82]: 
      2013-12-31  2014-12-31
BOB     0.163748    0.321372
ARON    0.471212    1.211516

sht.iloc[0:2,[0,1,3]]#前两行以及一二五列
Out[84]: 
      2012-12-31  2013-12-31  2015-12-31
BOB    -0.804521    0.163748    0.321230
ARON   -1.782169    0.471212   -0.002611

精确索引改变目标数值

sht.iloc[1,1]=10
sht
Out[98]: 
        2012-12-31  2013-12-31  2014-12-31  2015-12-31  2016-12-31
BOB      -0.804521    0.163748    0.321372    0.321230   -1.568735
ARON     -1.782169   10.000000    1.211516   -0.002611    1.971601
CARTER    0.490189    2.341276   -1.872003    0.422649    1.700321
DAVID     0.082172    1.387305   -0.142821   -0.616505    1.235456