Possible Duplicate:
Generating random results by weight in PHP?
I have a web application where users can add 1-20 strings of text and assign a weight to them of how often it should show up. The system would then choose a random string based on the defined weights. What is the best way to go about this? Do the range values for the weight for each string matter? Could I just have the user assign a number (0-100) for each string? How would you go about choosing a random string? (Each choice doesn't worry about what was chosen before, every string has the same odds (based on weight) of being chosen at the start of each call).
I use this function in several PHP game engines:
<?php
/**
* @param array $values - just the weights
* @return integer A number between 0 and count($values) - 1
*/
function getBucketFromWeights($values) {
$total = $currentTotal = $bucket = 0;
$firstRand = mt_rand(1, 100);
foreach ($values as $amount) {
$total += $amount;
}
$rand = ($firstRand / 100) * $total;
foreach ($values as $amount) {
$currentTotal += $amount;
if ($rand > $currentTotal) {
$bucket++;
}
else {
break;
}
}
return $bucket;
}
Usage
Suppose I have the user weights in an associative array where each string points to its weight:
$weighted_strings = array(
"important string" => 100,
"terrible string" => 10,
"never string" => 0,
// etc
);
If I wanted to pull a string based on weight, I'd do this:
$weights = array_values($weighted_strings);
$strings = array_keys($weighted_strings);
$index = getBucketFromWeights($weights);
$selectedString = $strings[$index];
Here is a simple implementation:
function Probability($data, $number = 1)
{
$result = array();
if (is_array($data) === true)
{
$data = array_map('abs', $data);
$number = min(max(1, abs($number)), count($data));
while ($number-- > 0)
{
$chance = 0;
$probability = mt_rand(1, array_sum($data));
foreach ($data as $key => $value)
{
$chance += $value;
if ($chance >= $probability)
{
$result[] = $key; unset($data[$key]); break;
}
}
}
}
return $result;
}
With this function you can specify how many unique weighted random elements you want (IDEOne).
精彩评论