对于merge、join和concat三者的区别分析

1、从合并的方式看merge和join是一样的，有left/right/inner/outer，而concat只有inner/outer两种，因为merge和join参与合并的对象有左右区分，而concat第一个参数是多个dataframe组成的列表，没有严格的左右区分，如果排除最后结果中列的顺序的话，可以看成是一样的。例：

import pandas as pd

from pandas import Series,DataFrame,Panel

df1 = DataFrame([['a','b'],['d','q'],['o','b'],['m','e']],index=['a','b','c','o'],columns=['number1','number2'])

df2 = DataFrame([['a','b'],['d','e'],['a','b'],['d','e']],columns=['col1','col2'],index=['f','a','g','c'])

print pd.concat([df1,df2],join='outer',axis=1)

print pd.concat([df2,df1],join='outer',axis=1)

2、merge合并的范围最广泛，可以合并左边对象的索引/列和右边对象的索引/列的四种组合；join次之，仅可以实现调用DataFrame的索引/列和参数DataFrame的索引的合并，也就是参数DataFrame的列不能参与合并；concat合并的范围最小，只支持索引的合并，也就是说索引与索引的合并是三个函数共同的功能，例：

import pandas as pd

import numpy as np

from pandas import Series,DataFrame,Panel

df1 = DataFrame([['a','b'],['d','q'],['o','b'],['m','e']],index=['a','b','c','o'],columns=['number1','number2'])

df2 = DataFrame([['a','b'],['d','e'],['a','b'],['d','e']],columns=['col1','col2'],index=['f','a','g','c'])

print(pd.merge(df1,df2,left_index=True,right_index=True,how='outer'))

print(df1.join(df2,how='outer'))

print(pd.concat([df1,df2],join='outer',axis=1))

可以看出三个函数合并索引后输出的结果是一样的

number1number2col1col2a a b d eb d q NaN NaNc o b d ef NaN NaN a bg NaN NaN a bo m e NaN NaN

number1number2col1col2a a b d eb d q NaN NaNc o b d ef NaN NaN a bg NaN NaN a bo m e NaN NaN

number1number2col1col2a a b d eb d q NaN NaNc o b d ef NaN NaN a bg NaN NaN a bo m e NaN NaN

3、concat和join的共同点是可以合并3个以上的对象，merge只合并两个对象