如果您需要通过以下方式进行多组：

    const populate = (entireObj, keys, item) => {
    let keysClone = [...keys],
        currentKey = keysClone.shift();

    if (keysClone.length > 0) {
        entireObj[item[currentKey]] = entireObj[item[currentKey]] || {}
        populate(entireObj[item[currentKey]], keysClone, item);
    } else {
        (entireObj[item[currentKey]] = entireObj[item[currentKey]] || []).push(item);
    }
}

export const groupBy = (list, key) => {
    return list.reduce(function (rv, x) {

        if (typeof key === 'string') (rv[x[key]] = rv[x[key]] || []).push(x);

        if (typeof key === 'object' && key.length) populate(rv, key, x);

        return rv;

    }, {});
}

const myPets = [
    {name: 'yaya', type: 'cat', color: 'gray'},
    {name: 'bingbang', type: 'cat', color: 'sliver'},
    {name: 'junior-bingbang', type: 'cat', color: 'sliver'},
    {name: 'jindou', type: 'cat', color: 'golden'},
    {name: 'dahuzi', type: 'dog', color: 'brown'},
];

// run 
groupBy(myPets, ['type', 'color']));

// you will get object like: 

const afterGroupBy = {
    "cat": {
        "gray": [
            {
                "name": "yaya",
                "type": "cat",
                "color": "gray"
            }
        ],
        "sliver": [
            {
                "name": "bingbang",
                "type": "cat",
                "color": "sliver"
            },
            {
                "name": "junior-bingbang",
                "type": "cat",
                "color": "sliver"
            }
        ],
        "golden": [
            {
                "name": "jindou",
                "type": "cat",
                "color": "golden"
            }
        ]
    },
    "dog": {
        "brown": [
            {
                "name": "dahuzi",
                "type": "dog",
                "color": "brown"
            }
        ]
    }
};

在我的特定用例中，我需要按属性分组，然后删除分组属性。

无论如何，该属性只是为了分组目的而添加到记录中的，对于向用户显示它没有意义。

    group (arr, key) {

        let prop;

        return arr.reduce(function(rv, x) {
            prop = x[key];
            delete x[key];
            (rv[prop] = (rv[prop] || [])).push(x);
            return rv;
        }, {});

    },

顶部答案中的起始函数归功于@caesar bautista。

您可以使用Alasql JavaScript库来实现：

var data = [ { Phase: "Phase 1", Step: "Step 1", Task: "Task 1", Value: "5" },
             { Phase: "Phase 1", Step: "Step 1", Task: "Task 2", Value: "10" }];

var res = alasql('SELECT Phase, Step, SUM(CAST([Value] AS INT)) AS [Value] \
                  FROM ? GROUP BY Phase, Step',[data]);

在jsFiddle尝试这个示例。

BTW：在大型阵列（100000条记录及以上）上，Alasql比Linq更快。参见jsPref中的测试。

评论：

这里我将Value放在方括号中，因为Value是SQL中的关键字我必须使用CAST（）函数将字符串值转换为数字类型。

data = [{id:1, name:'BMW'}, {id:2, name:'AN'}, {id:3, name:'BMW'}, {id:1, name:'NNN'}]
key = 'id'//try by id or name
data.reduce((previous, current)=>{
    previous[current[key]] && previous[current[key]].length != 0 ? previous[current[key]].push(current) : previous[current[key]] = new Array(current)
    return previous;
}, {})

基于@Ceasar Bautista的原始想法，我修改了代码并使用typescript创建了一个groupBy函数。

static groupBy(data: any[], comparator: (v1: any, v2: any) => boolean, onDublicate: (uniqueRow: any, dublicateRow: any) => void) {
    return data.reduce(function (reducedRows, currentlyReducedRow) {
      let processedRow = reducedRows.find(searchedRow => comparator(searchedRow, currentlyReducedRow));

      if (processedRow) {
        // currentlyReducedRow is a dublicateRow when processedRow is not null.
        onDublicate(processedRow, currentlyReducedRow)
      } else {
        // currentlyReducedRow is unique and must be pushed in the reducedRows collection.
        reducedRows.push(currentlyReducedRow);
      }

      return reducedRows;
    }, []);
  };

此函数接受一个回调（比较器）和一个第二个回调（onDuplicate），该回调比较行并查找副本。

用法示例：

data = [
    { name: 'a', value: 10 },
    { name: 'a', value: 11 },
    { name: 'a', value: 12 },
    { name: 'b', value: 20 },
    { name: 'b', value: 1 }
  ]

  private static demoComparator = (v1: any, v2: any) => {
    return v1['name'] === v2['name'];
  }

  private static demoOnDublicate = (uniqueRow, dublicateRow) => {
    uniqueRow['value'] += dublicateRow['value'];    
  };

使命感

groupBy(data, demoComparator, demoOnDublicate) 

将执行计算值和的分组。

{name: "a", value: 33}
{name: "b", value: 21}

我们可以根据项目的需要创建任意多个回调函数，并根据需要聚合这些值。在一个例子中，我需要合并两个数组，而不是求和数据。

让我们生成一个通用的Array.protocol.groupBy（）工具。为了多样化，让我们在递归方法上使用ES6 fancyty扩展运算符进行Haskell式模式匹配。同样，让我们让Array.prototype.groupBy（）接受一个回调，该回调将项（e）、索引（i）和应用的数组（a）作为参数。

Array.prototype.groupBy=函数（cb）{返回函数迭代（[x，…xs]，i=0，r=[[]，[]]）{cb（x，i，[x，…xs]）？（r[0].推（x），r）：（r[1].推（x），r）；是否返回xs.length？迭代（xs，++i，r）：r；}（本）；};var arr=[0,1,2,3,4,5,6,7,8,9]，res=arr.groupBy（e=>e<5）；console.log（res）；