GROUP BY 可以将别名中的数据行分成多个组，各组保留一行数据，其余的行舍弃；常与聚合、排序等操作配合使用。

语法：GROUP BY <expression> as <alias>, <expression> as <alias>, ...
参数：

• <expression>：分组依据，多个依据必须同源，从左到右进行逐级分组
• <alias>：分组依据的别名，可省略

例如，对一步路径 path 进行多重分组，先按照起点 n1 的形状分组，再在组内按照终点 n1 的颜色分组；统计各组的路径数量:

n(as n1).re().n(as n2) as path
group by n1.shape, n2.color
return path, count(path)

示例图集：(以下示例将在本图基础上运行)

create().node_schema("country").node_schema("movie").node_schema("director").edge_schema("filmedIn").edge_schema("direct")
create().node_property(@*, "name")
insert().into(@country).nodes([{_id:"C001", _uuid:1, name:"France"}, {_id:"C002", _uuid:2, name:"USA"}])
insert().into(@movie).nodes([{_id:"M001", _uuid:3, name:"Léon"}, {_id:"M002", _uuid:4, name:"The Terminator"}, {_id:"M003", _uuid:5, name:"Avatar"}])
insert().into(@director).nodes([{_id:"D001", _uuid:6, name:"Luc Besson"}, {_id:"D002", _uuid:7, name:"James Cameron"}])
insert().into(@filmedIn).edges([{_uuid:1, _from_uuid:3, _to_uuid:1}, {_uuid:2, _from_uuid:4, _to_uuid:1}, {_uuid:3, _from_uuid:3, _to_uuid:2}, {_uuid:4, _from_uuid:4, _to_uuid:2}, {_uuid:5, _from_uuid:5, _to_uuid:2}])
insert().into(@direct).edges([{_uuid:6, _from_uuid:6, _to_uuid:3}, {_uuid:7, _from_uuid:7, _to_uuid:4}, {_uuid:8, _from_uuid:7, _to_uuid:5}])

分组、组内聚合

示例：查找 2 步路径 国家-电影-导演，按导演分组，统计每组的路径条数

n({@country}).e().n({@movie}).e().n({@director} as n)
group by n
return table(n.name, count(n))

|     n.name    | count(n) |
|---------------|----------|
| Luc Besson    | 2        |
| James Cameron | 3        |

分析：聚合运算如果没有紧跟着 GROUP BY 子句，则会执行分组后的聚合，而非在组内进行聚合。

多重分组

示例：查找 2 步路径 国家-电影-导演，按国家-导演分组，统计每组的路径条数

n({@country} as a).e().n({@movie}).e().n({@director} as b)
group by a, b
return table(a.name, b.name, count(a))

| a.name |     b.name    | count(a) |
|--------|---------------|----------|
| France | Luc Besson    | 1        |
| France | James Cameron | 1        |
| USA    | Luc Besson    | 1        |
| USA    | James Cameron | 2        |