A PHP classifier in under 50 lines!

simple-naive-bayes-classifier.php

<?php 
function tokenize($string)
{
	$string = preg_replace("/ +/", " ", $string);
	return explode(" ", strtolower($string));
}

function train($dataset, $text, $class)
{
	$tokens=tokenize($text);
	
	$dataset['classes'][$class] = (empty($dataset['classes'][$class]) ? 0 : $dataset['classes'][$class]) + 1;

	foreach ($tokens as $token) {
		if(empty($dataset['tokens'][$token])){
			$dataset['tokens'][$token] = [];
		}
		$dataset['tokens'][$token][$class] = (empty($dataset['tokens'][$token][$class]) ? 0 : $dataset['tokens'][$token][$class]) + 1;
	}
	return $dataset;
}

function classify($dataset, $text){
	$sum_classes = 
	$tokens = tokenize($text);
	$probClasses = [];
	foreach(array_keys($dataset['classes']) as $class){
		$probTokens = [];
		foreach ($tokens as $token) {
			if(!empty($dataset['tokens'][$token])){
				if(empty($dataset['tokens'][$token][$class])){
					$probTokens[$token] = 0.000000001;
				} else {
					$probTokens[$token] = $dataset['tokens'][$token][$class] / $dataset['classes'][$class];
				}
			}
		}
		if(!empty($probTokens)){
			$probClasses[$class] = array_reduce(array_values($probTokens), function($a,$b){return $a*$b;},1);
		} else {
			$probClasses[$class] = 0;
		}
		$probClasses[$class] *= ($dataset['classes'][$class] / array_sum(array_values($dataset['classes'])));
	}
	array_multisort(array_keys($probClasses), SORT_DESC, $probClasses);
	return $probClasses;
}


$dataset = ['tokens' => [], 'classes' => []];
/*
$dataset = train($dataset, 'not to eat too much is not enough to lose weight', 'health');
$dataset = train($dataset, 'Russia try to invade Ukraine', 'politics');
$dataset = train($dataset, 'do not neglect exercise', 'health');
$dataset = train($dataset, 'Syria is the main issue, Obama says', 'politics');
$dataset = train($dataset, 'eat to lose weight', 'health');
$dataset = train($dataset, 'you should not eat much', 'health');

$classification = classify($dataset, "Obama is");
*/
die(var_dump($classification));