pandas 2

 

==============

sdf={'rkey':[1,2,3,2],'name':['rkey1','rkey2','rkey3','rkey4']}
sdf2={'lkey':[1,2,3],'name':['lsdf1','lsdf2','lsdf3']}
sdf3={'lkey':[11,2,3],'name':['lsdf3','lsdf3','lsdf3']}
cc=DataFrame(sdf)
cc2=DataFrame(sdf2)
cc3=DataFrame(sdf3)
print cc.merge(cc2,left_on='rkey',right_on='lkey')
name_x rkey lkey name_y

0 rkey1 1 1 lsdf1
1 rkey2 2 2 lsdf2
2 rkey4 2 2 lsdf2
3 rkey3 3 3 lsdf3

print cc2.merge(cc3,on='lkey')

lkey name_x name_y
0 2 lsdf2 lsdf3
1 3 lsdf3 lsdf3

# 指定附加在重名列上的字符串
print cc2.merge(cc3,on='lkey',suffixes=('_left','_right'))

lkey name_left name_right
0 2 lsdf2 lsdf3
1 3 lsdf3 lsdf3

 

======通过索引和列进行合并=====================

sdf2={'tst':[1,2,3],'name':['lsdf2','lsdf22','lsdf32']}
sdf3={'lkey':[11,2,3],'name':['lsdf3','lsdf23','lsdf33']}
cc2=DataFrame(sdf2,index=[1,2,3])
cc3=DataFrame(sdf3)

print cc2.merge(cc3,left_index=True,right_on='lkey')

 =======================

sdf=Series([11,22,33])
sdf2=Series([44,55,66])
print pd.concat([sdf,sdf2])

print pd.concat([sdf,sdf2],axis=1)

=============

sdf2={'tst2':[1,2,3],'name':['lsdf2','lsdf22','lsdf32']}
sdf3={'tst3':[11,2,3],'name':['lsdf3','lsdf23','lsdf33']}
cc2=DataFrame(sdf2,index=['b','c','d'])
cc3=DataFrame(sdf3,index=['a','b','c'])
print pd.concat([cc2,cc3])

print pd.concat([cc2,cc3],axis=1)

===========

print pd.concat([cc2,cc3],axis=1,join='inner')

print pd.concat([cc2,cc3],join='inner')

 ==============

sdf2={'tst':[1,2,3],'name':['lsdf2','lsdf22','lsdf32']}
sdf3={'tst':[11,2,3],'name':['lsdf3','lsdf23','lsdf33']}
cc2=DataFrame(sdf2,index=['b','c','d'])
cc3=DataFrame(sdf3,index=['a','b','c'])

print pd.concat([cc2,cc3],ignore_index=True)

 =========用参数对象中的数据为调用者对象的缺失数据打补丁==========

sdf2={'tst':[11,np.nan,33],'name':[np.nan,'lsdf22','lsdf22']}
sdf3={'tst':[1,2,3],'name':['lsdf3','lsdf23','lsdf33']}
cc2=DataFrame(sdf2,index=['b','c','d'])
cc3=DataFrame(sdf3,index=['a','b','c'])

print cc2.combine_first(cc3)

 ======================

sdf3={'tst':[1,2,3],'name':['lsdf3','lsdf23','lsdf33']}
cc3=DataFrame(sdf3,index=['a','b','c'])
# 指定附加在重名列上的字符串
print cc3.replace(3,100) #替换一个值

print cc3.replace([1,3],100) #替换多个值

print cc3.replace({1:100,3:300})  #不同值进行不同替换

 ================

df=pd.DataFrame({'name':['aa','bb','cc'],'age':[11,22,33]})
ss=df['age']
print ss

0    11
1    22
2    33
Name: age, dtype: int64

索引ss的某一个值：ss[0]

索引ss的某几个值：ss[[0,1]]

切片：ss[1:]

==========

s6=pd.Series(np.array([10,15,20,30,55,80]),index=['a','b','c','d','e','f'])
s7=pd.Series(np.array([12,11,13,15,14,16]),index=['a','c','g','b','d','f'])
#s6中不存在g索引，s7中不存在e索引，所以数据运算会产生两个缺失值NaN。
print(s6+s7)  
dtype: int32
a    22.0
b    30.0
c    31.0
d    44.0
e     NaN
f    96.0
g     NaN
#可以注意到这里的算术运算自动实现了两个序列的自动对齐
#对于数据框的对齐，不仅是行索引的自动对齐，同时也会对列索引进行自动对齐，数据框相当于二维数组的推广
print(s6/s7)
dtype: float64
a    0.833333
b    1.000000
c    1.818182
d    2.142857
e         NaN
f    5.000000
g         NaN
dtype: float64

获取DataFrame的多行：test_data.iloc[[0,2,4,5,7]]

按某一列的值进行过滤：test_data[test_data['age']==51]

对多列进行过滤：test_data[(test_data['age']==51) & (test_data['job']>=5)]   ---圆括号括起来+ &

过滤完后，只留下某几列：test_data[(test_data['age']==51) & (test_data['job']>=5)][['education','housing','loan','contact','poutcome']]

查询指定的行：test_data.loc[[0,2,4,5,7]]

查询指定的列：test_data[['age','job','marital']]

查询指定的行和列：test_data.loc[[0,2,4,5,7],['age','job','marital']]