se/SE/se-lib/Benford.php

<?php

class Benford {
	private $array = null;
	private $benford = null;

	public function __construct($array = null) {
		if ($array !== null) $this->addArray($array);
	}

	public function addArray($array = null) {
		if ($array === null) throw new Exception("Missing parameter");
		if (!is_array($array)) throw new Exception("Bad parameter - non array");
		foreach ($array as $k => $v) $this->addItem($k, $v);
	}

	public function addItem($k = null, $v = null) {
		if (!is_int($k)) throw new Exception("Bad key value - non integer ({$k})");
		if ($k < 1) throw new Exception("Bad key value - less than 1 ({$k})");
		if (isset($this->array[$k])) throw new Exception("Key value already exists ({$k})");
		if (!is_numeric($v)) throw new Exception("Bad value - non numeric ({$v})");
		$v = round(abs($v * 100));
		if ($v == 0) return;
		$this->array[$k] = $v;
		$this->benford = null;
	}

	public static function benford($k = null) {
		if ($k === null) throw new Exception("Missing parameter");
		if (!is_int($k)) throw new Exception("Bad parameter - non integer ({$k})");
		if ($k < 1) throw new Exceptrion("Bad parameter - less than 1 ({$k})");
		return log10(1 + 1 / $k);
	}

	public function calculate($secondStep = true) {
		return $this->calculateFirstStep($secondStep);
	}

	public function calculateFirstStep($secondStep = false) {
		if ($this->array === null) return false;
		$this->benford = [];

		foreach ($this->array as $k => $v) {
			$s = (string) $v;
			$this->benford['firstStep'][$s[0]]['keys'][] = $k;
		}

		for ($k = 1; $k <= 9; $k++) {
			if (!isset($this->benford['firstStep'][$k])) $this->benford['firstStep'][$k]['keys'] = [];
			$v = $this->benford['firstStep'][$k];
			$this->benford['firstStep'][$k]['count'] = count($v['keys']);
			$this->benford['firstStep'][$k]['value'] = count($v['keys']) / count($this->array);
			$this->benford['firstStep'][$k]['norm'] = self::benford($k);
			if ($this->benford['firstStep'][$k]['correct'] = ($this->benford['firstStep'][$k]['value'] <= $this->benford['firstStep'][$k]['norm'])) {
				unset($this->benford['firstStep'][$k]['keys']);
			} elseif ($secondStep) {
				$this->calculateSecondStep($k);
			}
		}
		ksort($this->benford['firstStep']);

		return true;
	}

	public function calculateSecondStep($a = null) {
		if ($a === null) {
			if (!isset($this->benford['firstStep'])) $this->calculate(false);
			foreach ($this->benford['firstStep'] as $k => $v) {
				if (!$v['correct']) $this->_calculateSecondStep($k);
			}
		} else $this->_calculateSecondStep($a);
	}

	private function _calculateSecondStep($a) {
		if (!isset($this->benford['firstStep'][$a]['keys'])) throw new Exception("Wrong data ({$a})");
		foreach ($this->benford['firstStep'][$a]['keys'] as $k) {
			$s = (string) $this->array[$k];
			if (strlen($s) < 2) continue;
			$this->benford['secondStep'][$a][$s[1]]['keys'][] = $k;
		}

		for ($k = 0; $k <= 9; $k++) {
			if (!isset($this->benford['secondStep'][$a][$k])) $this->benford['secondStep'][$a][$k]['keys'] = [];
			$v = $this->benford['secondStep'][$a][$k];
			$this->benford['secondStep'][$a][$k]['count'] = count($v['keys']);
			$this->benford['secondStep'][$a][$k]['value'] = count($v['keys']) / count($this->array);
			$this->benford['secondStep'][$a][$k]['norm'] = self::benford($a * 10 + $k);
			if ($this->benford['secondStep'][$a][$k]['correct'] = ($this->benford['secondStep'][$a][$k]['value'] <= $this->benford['secondStep'][$a][$k]['norm'])) {
				unset($this->benford['secondStep'][$a][$k]['keys']);
			}
		}

		ksort($this->benford['secondStep'][$a]);
	}

	private function getResult($type = null, $k = null) {
		if ($type === null) throw new Exception("Wrong usage");
		if ($this->benford === null) $this->calculateFirstStep();
		switch ($type) {
			case 'all':
			case 'secondStep':
				if (!isset($this->benford['secondStep'])) $this->calculateSecondStep();
			case 'firstStep':
				if ($type == 'all') return $this->benford;
				else return $this->benford[$type];
			case 'secondStepItem':
				if ($k === null) throw new Exception("Missing parameter");
				if (!is_int($k)) throw new Exception("Bad parameter - not integer ({$k})");
				if ($k < 1 || $k > 9) throw new Exception("Bad parameter - not between 1 and 9 ({$k})");
				if (!isset($this->benford['secondStep'][$k])) throw new Exception("Wrong data");
				return $this->benford['secondStep'][$k];
			case 'badFirstStep':
				return array_keys(array_filter($this->benford['firstStep'], function($v) {
					if (!$v['correct']) return true;
				}));
			default: throw new Exception('Wrong usage');
		}
	}

	public function getAllResult() {
		return $this->getResult('all');
	}

	public function getFirstStep() {
		return $this->getResult('firstStep');
	}

	public function getSecondStep() {
		return $this->getResult('secondStep');
	}

	public function getSecondStepItem($k = null) {
		return $this->getResult('secondStepItem', $k);
	}

	public function getBadFirstStep() {
		return $this->getResult('badFirstStep');
	}
}