merge函数可以将两个数据框按照指定的方式连接在一起，如果把数据框看做数据库中的一张表时，merge的功能可以看作是对两个表join 使用方法：

merge(x, y, by = , by.x = , by.y = , all = , all.x = , all.y = ,
      sort = , suffixes = , incomparables = , ...)

其中的参数解释如下

Argument	Description	Default
x	数据框1
y	数据框2
by	连接的列名	intersect(names(x), names(y))
by.x	第一个数据框要连接的列名	by
by.y	第二个数据框要连接的列名	by
all	如果连接后没有这一行的记录，是否要包含此行	FALSE
all.x	如果连接后第一个数据框中没有这一行的记录，是否要包含此行	all
all.y	如果连接后第二个数据框中没有这一行的记录，是否要包含此行	all

缺省下, merge等价于在两个表中的NATURAL JOIN，或者设置by的值使得merge等价于INNER JOIN，也可以设置all的值来使得merge等价于OUTER JOIN或者FULL，在缺省条件下，如果在两个数据框中没有相同的列名，则使用CROSS JOIN 比如有如下两个数据框

> persons
  id_p firstname lastname       address     city
1    1      John    Adams Oxford Street   London
2    2    George     Bush  Fifth Avenue New York
3    3    Thomas   Carter Changa Street  Beijing
> orders
  id_o orderno id_p
1    1   77985    3
2    2   44678    3
3    3   22456    1
4    4   24562    1
5    5   34764   65

因为其中有同名的列名，使用merge而不指定要连接的列名

> merge(orders, persons)
  id_p id_o orderno firstname lastname       address    city
1    1    3   22456      John    Adams Oxford Street  London
2    1    4   24562      John    Adams Oxford Street  London
3    3    1   77985    Thomas   Carter Changa Street Beijing
4    3    2   44678    Thomas   Carter Changa Street Beijing

这个如下的内连接相同

> merge(orders, persons, by="id_p")
> merge(orders, persons, by.x="id_p", by.y="id_p")

等价于如下的SQL语句

SELECT o.id_p,
       o.id_o,
       o.orderno,
       p.firstname,
       p.lastname,
       p.address,
       p.city
FROM orders AS o
     AND persons AS p
WHERE o.id_p = op.id_p

> merge(orders, persons, by.x = "id_p", by.y = "id_p", all.x = TRUE)
  id_p id_o orderno firstname lastname       address    city
1    1    3   22456      John    Adams Oxford Street  London
2    1    4   24562      John    Adams Oxford Street  London
3    3    1   77985    Thomas   Carter Changa Street Beijing
4    3    2   44678    Thomas   Carter Changa Street Beijing
5   65    5   34764      <NA>     <NA>          <NA>    <NA>
> merge(persons, orders, by.x = "id_p", by.y = "id_p", all.y = TRUE)

等价于

SELECT o.id_p,
       o.id_o,
       o.orderno,
       p.firstname,
       p.lastname,
       p.address,
       p.city
FROM orders AS o
     LEFT JOIN persons AS p
     ON o.id_p = op.id_p

SELECT o.id_p,
       o.id_o,
       o.orderno,
       p.firstname,
       p.lastname,
       p.address,
       p.city
FROM persons AS p
     RIGHT JOIN orders AS o
     ON o.id_p = op.id_p

> merge(orders, persons, by.x = "id_p", by.y = "id_p", all = TRUE)
  id_p id_o orderno firstname lastname       address     city
1    1    3   22456      John    Adams Oxford Street   London
2    1    4   24562      John    Adams Oxford Street   London
3    2   NA      NA    George     Bush  Fifth Avenue New York
4    3    1   77985    Thomas   Carter Changa Street  Beijing
5    3    2   44678    Thomas   Carter Changa Street  Beijing
6   65    5   34764      <NA>     <NA>          <NA>     <NA>

等价于

SELECT o.id_p,
       o.id_o,
       o.orderno,
       p.firstname,
       p.lastname,
       p.address,
       p.city
FROM orders AS o
     OUTER JOIN persons AS p