Gradient Descent

gradient_descent.php

<?php

function computeErrorForLineGivenPoints($b, $m, array $points)
{
    $total_error = 0;

    foreach ($points as $point)
    {
        $total_error += pow(($point['y'] - ($m * $point['x'] + $b)), 2);
    }

    return $total_error/count($points);
}

function stepGradient($b_current, $m_current, array $points, $learning_rate)
{
    $b_gradient = 0;
    $m_gradient = 0;

    $n = (float)count($points);

    foreach ($points as $point)
    {
        $b_gradient += -(2 / $n) * ($point['y'] - (($m_current * $point['x']) + $b_current));
        $m_gradient += -(2 / $n) * $point['x'] * ($point['y'] - (($m_current * $point['x']) + $b_current));
    }

    $new_b = $b_current - ($learning_rate * $b_gradient);
    $new_m = $m_current - ($learning_rate * $m_gradient);
    
    return [$new_b, $new_m];
}

Functions for performing linear regression via gradient descent.

Ported from python samples seen here:
http://spin.atomicobject.com/2014/06/24/gradient-descent-linear-regression/

I'm mixing platforms too - I wrote a Python based code generator to write Laravel (PHP) code. It's a bit rough around the edges, but it does a decent job.