1、從合并的方式看merge和join是一樣的，有left/right/inner/outer，而concat只有inner/outer兩種，因為merge和join參與合并的對象有左右區分，而concat第一個參數是多個dataframe組成的列表，沒有嚴格的左右區分，如果排除最后結果中列的順序的話，可以看成是一樣的。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合并的范圍最廣泛，可以合并左邊對象的索引/列和右邊對象的索引/列的四種組合；join次之，僅可以實現調用DataFrame的索引/列和參數DataFrame的索引的合并，也就是參數DataFrame的列不能參與合并；concat合并的范圍最小，只支持索引的合并，也就是說索引與索引的合并是三個函數共同的功能，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))

可以看出三個函數合并索引后輸出的結果是一樣的

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的共同點是可以合并3個以上的對象，merge只合并兩個對象