简体中文（ZH-CN）English（EN）繁体中文（ZH-TW）日本語（JA）한국어（KO）Melayu（MS）Français（FR）Deutsch（DE）
Home
Article
Q&A
Course
Topic
Download
Game
Dictionary
Recent Updates

Home  >  Article  >  Web Front-end  >  Introduction to the weight probability allocation method of js array

Introduction to the weight probability allocation method of js array

巴扎黑
巴扎黑Original
2017-09-13 09:19:561559browse

Today I wrote a js function to control page carousel. It is very simple if you just use the queue, but considering whether assigning weight to each page becomes extremely complicated, it cannot be solved using switch and if else, so I thought of using js array implementation

Today I wrote a js function to control page carousel. It is very simple if you just use the queue, but considering whether assigning weight to each page becomes extremely complicated, using switch and if else is also Unable to solve it, I thought of using a js array to implement it. The idea is to abstract each carousel page into an object. Each object needs to manually specify the weight value, and then form an array. Using the function encapsulated below, it will be based on the corresponding weight of each object. Probability returns an object, the code is as follows: 


/**
* js数组实现权重概率分配
* @param  Array  arr    js数组，参数类型[Object,Object,Object……]
* @return  Array        返回一个随机元素，概率为其percent/所有percent之和，参数类型Object
* @author  shuiguang
*/
function weight_rand(arr){
  //参数arr元素必须含有percent属性，参考如下所示
  /*
  var arr = [{
      name : '1',
      percent : 1
    }, {
      name : '2',
      percent : 2
    }, {
      name : '3',
      percent : 1
    }, {
      name : '4',
      percent : 2
    }
  ];
  */
  var total = 0;
  var i, j, percent;
  //下标标记数组，按照上面的例子，单倍情况下其组成为[1,2,2,3,4,4]
  var index = new Array();
  for (i = 0; i < arr.length; i++) {
    //判断元素的权重，为了实现小数权重，先将所有的值放大100倍
    percent = &#39;undefined&#39; != typeof(arr[i].percent) ? parseInt(arr[i].percent*100) : 0;
    for (j = 0; j < percent; j++) {
      index.push(i);
    }
    total += percent;
  }
  //随机数值，其值介于0-5的整数
  var rand = Math.floor(Math.random() * total);
  return arr[index[rand]];
}

Although the above method is feasible, it encounters such a problem: for general complex allocation situations such as 1:1:1 allocation (relative value) can be satisfied, but if you encounter the exact weight distribution (absolute value) of 15%, 25%, 35% remaining, etc., it cannot be satisfied. Because it is very troublesome to calculate the remaining ratio of 15%:25%:35%, I continued to modify the above function and added a percentage mode. For example, in the above example, after allocating the clear percentages above, the remaining percentages will be Give the last element without calculating the percentage of the last element or the proportion of each element. The code is as follows: 


/**
* js数组实现权重概率分配，支持数字比模式(支持2位小数)和百分比模式(不支持小数，最后一个元素多退少补)
* @param  Array  arr  js数组，参数类型[Object,Object,Object……]
* @return  Array      返回一个随机元素，概率为其weight/所有weight之和，参数类型Object
* @author  shuiguang
*/
function weight_rand(arr){
	//参数arr元素必须含有weight属性，参考如下所示
	//var arr=[{name:&#39;1&#39;,weight:1.5},{name:&#39;2&#39;,weight:2.5},{name:&#39;3&#39;,weight:3.5}];
	//var arr=[{name:&#39;1&#39;,weight:&#39;15%&#39;},{name:&#39;2&#39;,weight:&#39;25%&#39;},{name:&#39;3&#39;,weight:&#39;35%&#39;}];
	//求出最大公约数以计算缩小倍数，perMode为百分比模式
	var per;
	var maxNum = 0;
	var perMode = false;
	//自定义Math求最小公约数方法
	Math.gcd = function(a,b){
		var min = Math.min(a,b);
		var max = Math.max(a,b);
		var result = 1;
		if(a === 0 || b===0){
			return max;
		}
		for(var i=min; i>=1; i--){
			if(min % i === 0 && max % i === 0){
				result = i;
				break;
			}
		}
		return result;
	};
	
	//使用clone元素对象拷贝仍然会造成浪费，但是使用权重数组对应关系更省内存
	var weight_arr = new Array();
	for (i = 0; i < arr.length; i++) {
		if(&#39;undefined&#39; != typeof(arr[i].weight))
		{
			if(arr[i].weight.toString().indexOf(&#39;%&#39;) !== -1) {
				per = Math.floor(arr[i].weight.toString().replace(&#39;%&#39;,&#39;&#39;));
				perMode = true;
			}else{
				per = Math.floor(arr[i].weight*100);
			}
		}else{
			per = 0;
		}
		weight_arr[i] = per;
		maxNum = Math.gcd(maxNum, per);
	}
	//数字比模式，3:5:7，其组成[0, 0, 0, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2]
	//百分比模式，元素所占百分比为15%，25%，35%
	var index = new Array();
	var total = 0;
	var len = 0;
	if(perMode){
		for (i = 0; i < arr.length; i++) {
			//len表示存储arr下标的数据块长度，已优化至最小整数形式减小索引数组的长度
			len = weight_arr[i];
			for (j = 0; j < len; j++){
				//超过100%跳出，后面的舍弃
				if(total >= 100){
					break;
				}
				index.push(i);
				total++;
			}
		}
		//使用最后一个元素补齐100%
		while(total < 100){
			index.push(arr.length-1);
			total++;
		}
	}else{
		for (i = 0; i < arr.length; i++) {
			//len表示存储arr下标的数据块长度，已优化至最小整数形式减小索引数组的长度
			len = weight_arr[i]/maxNum;
			for (j = 0; j < len; j++){
				index.push(i);
			}
			total += len;
		}
	}
	//随机数值，其值为0-11的整数，数据块根据权重分块
	var rand = Math.floor(Math.random()*total);
	//console.log(index);
	return arr[index[rand]];
}

var arr=[{name:&#39;1&#39;,weight:1.5},{name:&#39;2&#39;,weight:2.5},{name:&#39;3&#39;,weight:3.5}];
console.log(weight_rand(arr));
var arr=[{name:&#39;1&#39;,weight:&#39;15%&#39;},{name:&#39;2&#39;,weight:&#39;25%&#39;},{name:&#39;3&#39;,weight:&#39;35%&#39;}];
console.log(weight_rand(arr));
var prize_arr = [
	{&#39;id&#39;:1, &#39;prize&#39;:&#39;平板电脑&#39;, &#39;weight&#39;:1},
	{&#39;id&#39;:2, &#39;prize&#39;:&#39;数码相机&#39;, &#39;weight&#39;:2},
	{&#39;id&#39;:3, &#39;prize&#39;:&#39;音箱设备&#39;, &#39;weight&#39;:10},
	{&#39;id&#39;:4, &#39;prize&#39;:&#39;4G优盘&#39;, &#39;weight&#39;:12},
	{&#39;id&#39;:5, &#39;prize&#39;:&#39;10Q币&#39;, &#39;weight&#39;:22},
	{&#39;id&#39;:6, &#39;prize&#39;:&#39;下次没准就能中哦&#39;, &#39;weight&#39;:50}    
];

var times = 100000;
var prize;
var pingban = 0;
var shuma = 0;
var yinxiang = 0;
var youpan = 0;
var qb = 0;
var xc = 0;
var start = new Date().getTime();

for($i=0; $i<times; $i++){
	prize = weight_rand(prize_arr);
	if(prize.prize == &#39;平板电脑&#39;)
	{
		pingban++;
	}else if(prize.prize == &#39;数码相机&#39;){
		shuma++;
	}else if(prize.prize == &#39;音箱设备&#39;){
		yinxiang++;
	}else if(prize.prize == &#39;4G优盘&#39;){
		youpan++;
	}else if(prize.prize == &#39;10Q币&#39;){
		qb++;
	}else if(prize.prize == &#39;下次没准就能中哦&#39;){
		xc++;
	}
}

var stop = new Date().getTime();
console.log(&#39;平板电脑：&#39;+pingban/times+&#39;, 数码相机：&#39;+shuma/times+&#39;, 音箱设备：&#39;+yinxiang/times+&#39;, 4G优盘：&#39;+youpan/times+&#39;, 10Q币：&#39;+qb/times+&#39;, 下次没准就能中哦：&#39;+xc/times);
console.log(&#39;耗费时间：&#39;+(stop-start)/1000+&#39;秒&#39;);

The code has been optimized for the subscript array through the greatest common divisor. The number ratio mode has been optimized to the minimum numerical ratio. The percentage mode is considered Performance consumption does not currently support 2 decimal places.

After writing the js version, I easily changed to the php version. After 100,000 loop tests, I found that the for loop saves time than foreach, and the foreach uploaded off the Internet is faster than for. But in general, the execution speed of js is about 20 times that of php. The execution time of php is about 6 seconds, and the execution time of js is about 0.346 seconds. 


/**
* php数组实现权重概率分配，支持数字比模式(支持2位小数)和百分比模式(不支持小数，最后一个元素多退少补)
* @param  array  $arr  php数组，参数类型array(array(),array(),array()……)
* @return  array      返回一个随机元素，概率为其percent/所有percent之和，参数类型array()
* @author  shuiguang
*/
function weight_rand($arr)
{
  //参数arr元素必须含有percent属性，参考如下所示
  //$arr=array(array(&#39;name&#39;=>&#39;1&#39;,&#39;weight&#39;=>1.5),array(&#39;name&#39;=>&#39;2&#39;,&#39;weight&#39;=>1.5),array(&#39;name&#39;=>&#39;3&#39;,&#39;weight&#39;=>1.5));
  //$arr=array(array(&#39;name&#39;=>&#39;1&#39;,&#39;weight&#39;=>&#39;15%&#39;),array(&#39;name&#39;=>&#39;2&#39;,&#39;weight&#39;=>&#39;25%&#39;),array(&#39;name&#39;=>&#39;3&#39;,&#39;weight&#39;=>&#39;35%&#39;));
  //求出最大公约数以计算缩小倍数，perMode为百分比模式
  $perMode = false;
  $maxNum = 0;
  //自定义求最小公约数方法
  $gcd = function($a, $b)
  {
    $min = min($a, $b);
    $max = max($a, $b);
    $result = 1;
    if($a === 0 || $b === 0)
    {
      return $max;
    }
    for($i=$min; $i>=1; $i--)
    {
      if($min % $i === 0 && $max % $i === 0)
      {
        $result = $i;
        break;
      }
    }
    return $result;
  };
  //使用传地址可能会影响后面的结果，但是使用权重数组对应关系更省内存
  $weight_arr = array();
  $arr_len = count($arr);
  for($i=0; $i<$arr_len; $i++)
  {
    if(isset($arr[$i][&#39;weight&#39;]))
    {
      if(strpos($arr[$i][&#39;weight&#39;], &#39;%&#39;) !== false)
      {
        $per = floor(str_replace(&#39;%&#39;, &#39;&#39;, $arr[$i][&#39;weight&#39;]));
        $perMode = true;
      }else{
        $per = floor($arr[$i][&#39;weight&#39;]*100);
      }
    }else{
      $per = 0;
    }
    $weight_arr[$i] = $per;
    $maxNum = call_user_func($gcd, $maxNum, $per);
  }
  //数字比模式，3:5:7，其组成[0, 0, 0, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2]
  //百分比模式，元素所占百分比为15%，25%，35%
  $index = array();
  $total = 0;
  if($perMode)
  {
    for($i=0; $i<$arr_len; $i++)
    {
      //$len表示存储$arr下标的数据块长度，已优化至最小整数形式减小索引数组的长度
      $len = $weight_arr[$i];
      for ($j = 0; $j < $len; $j++)
      {
        //超过100%跳出，后面的舍弃
        if($total >= 100)
        {
          break;
        }
        $index[] = $i;
        $total++;
      }
    }
    //使用最后一个元素补齐100%
    while($total < 100)
    {
      $index[] = $arr_len-1;
      $total++;
    }
  }else{
    for($i=0; $i<$arr_len; $i++)
    {
      //len表示存储arr下标的数据块长度，已优化至最小整数形式减小索引数组的长度
      $len = $weight_arr[$i]/$maxNum;
      for ($j = 0; $j < $len; $j++)
      {
        $index[] = $i;
      }
      $total += $len;
    }
  }
  //随机数值，其值为0-11的整数，数据块根据权重分块
  $rand = floor(mt_rand(0, $total));
	//修复php随机函数可以取临界值造成的bug
  $rand = $rand == $total ? $total-1 : $rand;
  return $arr[$index[$rand]];
}

$arr=array(array(&#39;name&#39;=>&#39;1&#39;,&#39;weight&#39;=>1.5),array(&#39;name&#39;=>&#39;2&#39;,&#39;weight&#39;=>1.5),array(&#39;name&#39;=>&#39;3&#39;,&#39;weight&#39;=>1.5));
p(weight_rand($arr));
$arr=array(array(&#39;name&#39;=>&#39;1&#39;,&#39;weight&#39;=>&#39;15%&#39;),array(&#39;name&#39;=>&#39;2&#39;,&#39;weight&#39;=>&#39;25%&#39;),array(&#39;name&#39;=>&#39;3&#39;,&#39;weight&#39;=>&#39;35%&#39;));
p(weight_rand($arr));

$prize_arr = array(
	&#39;0&#39; => array(&#39;id&#39;=>1, &#39;prize&#39;=>&#39;平板电脑&#39;, &#39;weight&#39;=>1),
	&#39;1&#39; => array(&#39;id&#39;=>2, &#39;prize&#39;=>&#39;数码相机&#39;, &#39;weight&#39;=>5),
	&#39;2&#39; => array(&#39;id&#39;=>3, &#39;prize&#39;=>&#39;音箱设备&#39;, &#39;weight&#39;=>10),
	&#39;3&#39; => array(&#39;id&#39;=>4, &#39;prize&#39;=>&#39;4G优盘&#39;, &#39;weight&#39;=>12),
	&#39;4&#39; => array(&#39;id&#39;=>5, &#39;prize&#39;=>&#39;10Q币&#39;, &#39;weight&#39;=>22),
	&#39;5&#39; => array(&#39;id&#39;=>6, &#39;prize&#39;=>&#39;下次没准就能中哦&#39;, &#39;weight&#39;=>50),
);

$start = time();
$result = array();
$times = 100000;
for($i=0; $i<$times; $i++)
{
	$row = weight_rand($prize_arr);
	if(array_key_exists($row[&#39;prize&#39;], $result))
	{
		$result[$row[&#39;prize&#39;]] ++;
	}else{
		$result[$row[&#39;prize&#39;]] = 1;
	}
}
$cost = time() - $start;


p($result);
p(&#39;耗费时间：&#39;.$cost.&#39;秒&#39;);
function p($var)
{
  echo "<pre class="brush:php;toolbar:false">";
  if($var === false)
  {
    echo &#39;false&#39;;
  }else if($var === &#39;&#39;){
    print_r("&#39;&#39;");
  }else{
    print_r($var);
  }
  echo "
"; }

php version If you just use the integer ratio mode, you don’t need to consider the amplification of numbers and the algorithm for finding the least common multiple. You only need to do simple accumulation, which can greatly Reduce execution time.

The above is the detailed content of Introduction to the weight probability allocation method of js array. For more information, please follow other related articles on the PHP Chinese website!

Statement：
The content of this article is voluntarily contributed by netizens, and the copyright belongs to the original author. This site does not assume corresponding legal responsibility. If you find any content suspected of plagiarism or infringement, please contact admin@php.cn
Previous article：js realizes the occurrence of 1,2,3,5 according to a certain probabilityNext article：js realizes the occurrence of 1,2,3,5 according to a certain probability

Related articles

See more