How does this W3Schools code for shuffling an array with .sort() work?

前端 未结 2 1140
无人共我
无人共我 2021-01-21 19:34

This code snippet is from w3schools JavaScript section. I am trying to figure out what

points.sort( function(a, b) {
  return 0.5 - Math.random()
});

相关标签:
2条回答
  • 2021-01-21 19:46

    The sort callback is supposed to return a value <0, 0 or >0 to indicate whether the first value is lower than, equal to or higher than the second; sort uses that to sort the values. 0.5 - Math.random() returns a value between -0.5 and 0.5 randomly, satisfying the expected return values and resulting in an essentially randomly shuffled array.

    Note that this shouldn't be the preferred method to shuffle; since the return value is random and not internally consistent (e.g. it tells sort that foo < bar, bar < baz and foo > baz), it may make Javascript's sort algorithm very inefficient. A Fisher-Yates shuffle for instance is pretty trivially implemented and likely more efficient.

    0 讨论(0)
  • 2021-01-21 20:05

    Passing a callback to Array#sort() that returns a random value is a bad idea; the ECMAScript spec provides no guarantees about what will happen in that case. Per MDN:

    compareFunction(a, b) must always return the same value when given a specific pair of elements a and b as its two arguments. If inconsistent results are returned then the sort order is undefined.

    And per the spec itself:

    If comparefn is not undefined and is not a consistent comparison function for the elements of this array (see below), the sort order is implementation-defined.

    The W3Schools code from the question is demonstrably broken; it doesn't shuffle the array fairly, at least in Chrome. Let's try running it a million times and counting how often each value shows up in the final position in the array after "shuffling":

    function shuffleAndTakeLastElement() {
        var points = [40, 100, 1, 5, 25, 10];
        return points.sort(function(a, b){return 0.5 - Math.random()})[5];
    }
    results = {};
    for (var point of [40, 100, 1, 5, 25, 10]) results[point] = 0;
    for (var i = 0; i < 1000000; i++) results[shuffleAndTakeLastElement()]++;
    console.log(results);
    

    I get the following counts in Chrome:

    {1: 62622, 5: 125160, 10: 500667, 25: 249340, 40: 31057, 100: 31154}
    

    Notice how the number 10 is around 16 times more likely to end up in the end position of the array than the numbers 40 or 100 are. This ain't a fair shuffle!

    A few morals to draw from this story:

    • You should run a large number of tests and look at the results to help confirm whether any randomness algorithm is fair.
    • It's easy to accidentally write a biased algorithm even if you're starting with a fair source of randomness.
    • For shuffling arrays, use Lodash's _.shuffle method or one of the approaches from How to randomize (shuffle) a JavaScript array?.
    • Never trust anything you read on W3Schools, because they suck and are riddled with errors.
    0 讨论(0)
提交回复
热议问题