如何根据javascript中的权重值从数组中获取n个值

Question

我有一个这样的数组，其中包含大约 30 万个条目。

var data = [
 {name: "peter", city: "bauru", weightage: 20},
 {name: "marcos", city: "belo", weightage: 15},
 {name: "rio", city: "salvador", weightage: 20},
 {name: "halter", city: "natal", weightage: 15},
 {name: "ron", city: "belem", weightage: 15},
 {name: "dominic", city: "santos", weightage: 15},
 {name: "john", city: "bauru", weightage: 20},
 {name: "ros", city: "natal", weightage: 15},
 {name: "nick", city: "salvador", weightage: 20},
 {name: "david", city: "santos", weightage: 15},
 {name: "ison", city: "belem", weightage: 15},
 {name: "eddy", city: "belo", weightage: 15},
....
 
]

我在一个数组中有条目，每个不同的用户都位于这 6 个城市下。每个城市都有权重值。所以我要根据权重值随机挑选60000条数据。

我尝试首先分离数据并从 bauru 中选择 20% 的数据，从 belo 中选择 15% 的数据等等，但这种手动操作并不是一个好方法。

我不知道在 javascript 上写这个。

预期输出：我想要 60k 数据，其中 20% 的值来自 bauru，15% 来自贝洛，20% 来自萨尔瓦多，15% 来自 natal，15%来自贝伦，15% 来自桑托斯。总数为 60k。如果任何城市数据总计没有出现 15% 或 20%，则从 20% 和 15% 中选择更多数据。假设 60k 数据的 15% = 7200，任何城市的总值只有 500，那么我们从数据量大的不同城市的 20% 或 15% 中挑选其余数据

Answer 1

您可以先按城市分组，然后从组中选择项目的数量。

var data = [{ name: "peter", city: "bauru", weightage: 20 }, { name: "marcos", city: "belo", weightage: 15 }, { name: "rio", city: "salvador", weightage: 20 }, { name: "halter", city: "natal", weightage: 15 }, { name: "ron", city: "belem", weightage: 15 }, { name: "dominic", city: "santos", weightage: 15 }, { name: "john", city: "bauru", weightage: 20 }, { name: "ros", city: "natal", weightage: 15 }, { name: "nick", city: "salvador", weightage: 20 }, { name: "david", city: "santos", weightage: 15 }, { name: "ison", city: "belem", weightage: 15 }, { name: "eddy", city: "belo", weightage: 15 }, { name: "queen", city: "bauru", weightage: 20 }, { name: "lucia", city: "belo", weightage: 15 }, { name: "fran", city: "salvador", weightage: 20 }, { name: "yvonne", city: "natal", weightage: 15 }, { name: "audry", city: "belem", weightage: 15 }, { name: "gaby", city: "santos", weightage: 15 }, { name: "susan", city: "bauru", weightage: 20 }, { name: "tonya", city: "natal", weightage: 15 }, { name: "ines", city: "salvador", weightage: 20 }, { name: "karen", city: "santos", weightage: 15 }, { name: "petra", city: "belem", weightage: 15 }, { name: "jane", city: "belo", weightage: 15 }, { name: "jo", city: "bauru", weightage: 20 }, { name: "kim", city: "belo", weightage: 15 }, { name: "tony", city: "salvador", weightage: 20 }, { name: "adrian", city: "natal", weightage: 15 }, { name: "alex", city: "belem", weightage: 15 }, { name: "dana", city: "santos", weightage: 15 }, { name: "sam", city: "bauru", weightage: 20 }, { name: "nicola", city: "natal", weightage: 15 }, { name: "finn", city: "salvador", weightage: 20 }, { name: "sasha", city: "santos", weightage: 15 }, { name: "kay", city: "belem", weightage: 15 }, { name: "ira", city: "belo", weightage: 15 }],
    target = 20,
    groups = data.reduce((r, o) => {
        if (!r[o.city]) r[o.city] = [];
        r[o.city].push(o);
        return r;
    }, {}),
    result = Object.keys(groups).reduce((r, k) => {
        let count = Math.floor(target * groups[k][0].weightage/ 100);
        while (count-- && groups[k].length)
            r.push(groups[k].splice(Math.floor(Math.random() * groups[k].length), 1)[0]);
        return r;
    }, []);

console.log(result);

.as-console-wrapper { max-height: 100% !important; top: 0; }

Answer 2

此解决方案按城市对所有数据进行分组，随机化每组中记录的顺序，并从每组中获取前 X 条记录，其中 X 是达到所需权重所需的总数。

在样本数据中，总数为 3、4 和 3，而在您的实际数据中，大多数城市（权重 15）使用 9000，其他城市使用 12（权重 20）。

（上次我查看您的问题时，您尚未添加规范，即某些城市可能没有足够的数据来达到其所需的权重，因此未处理这种情况，但我也许可以添加一些内容如果您仍然需要帮助，请明天再做。）

    // Creates the `cities` object and retrieves data
    const
      cities = getCities(),
      data = getData();

    // Copies all records from `data` into cities, grouped by city
    data.reduce( (acc, item) =>
      addRecordToCity(item, item.city), cities);

    // Randomizes and then truncates each city-specific array
    Object.keys(cities).forEach(city => {
      let
        cityData = cities[city].data,
        numberNeeded = cities[city].total;
      cities[city].data = shuffle(cityData).slice(0, numberNeeded);
    });

    // Prints results
    console.log("Selections:")
    Object.keys(cities).forEach(city => {
      cities[city].data.forEach(item =>
        console.log(item.name, item.city)
      );
    });


    // Coptes a record from `data` into `cities`, in the appropriate city
    function addRecordToCity(record, cityName){
      cities[cityName].data.push(record);
      return cities;
    }

    // Randomizes an array
    function shuffle(array){
      let i = array.length, rand, temp;
      while(--i > 0){
        rand = Math.floor( Math.random() * (i + 1) );
        temp = array[rand];
        array[rand] = array[i]
        array[i] = temp;
      }
      return array;
    }

    // Defines the cities object that will hold our final results
    function getCities(){
      return {  // Weightages are 30/40/30, collecting 10 records total
        "city_1": { total: 3, data: [] },
        "city_2": { total: 4, data: [] },
        "city_3": { total: 3, data: [] }
      };
    }

    // Defines source data
    function getData(){
      return [
        { name: "a", city: "city_1" },
        { name: "b", city: "city_2" },
        { name: "c", city: "city_3" },
        { name: "d", city: "city_1" },
        { name: "e", city: "city_2" },
        { name: "f", city: "city_3" },
        { name: "g", city: "city_1" },
        { name: "h", city: "city_2" },
        { name: "i", city: "city_3" },
        { name: "j", city: "city_1" },
        { name: "k", city: "city_2" },
        { name: "l", city: "city_3" },
        { name: "m", city: "city_1" },
        { name: "n", city: "city_2" },
        { name: "o", city: "city_3" },
        { name: "p", city: "city_1" },
        { name: "q", city: "city_2" },
        { name: "r", city: "city_3" }
      ];
    }

Answer 3

如果您创建一个控制表来指示您希望如何选择样本，这应该相当简单。

我建议分两步进行。首先将您的完整记录集过滤为仅那些您认为“有效”的记录（即具有正确的城市/年龄组合），然后您可以在最终集中挑选您想要的人：

var data = [
 {name: "peter", city: "bauru", weightage: 20},
 {name: "marcos", city: "belo", weightage: 15},
 {name: "rio", city: "salvador", weightage: 20},
 {name: "halter", city: "natal", weightage: 15},
 {name: "ron", city: "belem", weightage: 15},
 {name: "dominic", city: "santos", weightage: 15},
 {name: "john", city: "bauru", weightage: 20},
 {name: "ros", city: "natal", weightage: 15},
 {name: "nick", city: "salvador", weightage: 20},
 {name: "david", city: "santos", weightage: 15},
 {name: "ison", city: "belem", weightage: 15},
 {name: "eddy", city: "belo", weightage: 15}
]

//create a control array with the matches you want (city/weightage etc) 
//include the result percent you care about
let controlArray = [
  {city: "bauru", weightage: 20, resultPercent : .2},
  {city: "belo", weightage: 15, resultPercent : .2},
  {city: "salvador", weightage: 20, resultPercent : 0},
  {city: "natal", weightage: 15, resultPercent : .2},
  {city: "belem", weightage: 15, resultPercent : .2},
  {city: "santos", weightage: 15, resultPercent : .2}
];

//this is your result set
let validRecords = [];

//this is your sample size - I've used 5 as the sample data was only 12 rows
let requiredRecords = 5;

//iterate over the control array
controlArray.forEach(controlRecord => {
    //available records - here you filter out records that don't match your criteria
    let availablePeople = data.filter(individual => {return individual.city == controlRecord.city && individual.weightage == controlRecord.weightage});
  
    //get a sample from the reduced set
    let requiredNumberFromSet = Math.ceil(controlRecord.resultPercent*requiredRecords);

    validRecords = validRecords.concat(availablePeople.slice(0, requiredNumberFromSet));
});

//your result set, here you should see five records - one from each city excluding salvador which I set to 0% for this demo.
console.log(validRecords);

如果一组没有足够的有效匹配项，您也可以相当容易地在此方法中添加“溢出”逻辑，您可以将其存储为变量并将其添加到下一个控制行中的选择中。