Good polynomial or linear regression script?

[virtual_odin]

Hey! It's been years since I did a stats course, but I know what these babies can do... And Excel has this built in! I now need one for a site I am building. If only I could find the code to generate the results from my dataset. I really do not fancy going back to school to re-learn how to construct them and then have to convert those principles to PHP. I'm easy as to whether it is PHP, taking a two dimensional array and producing the a, b (as in y = ax + b) and the correlation, or MySQL (I'm running version 5 and I saw somewhere that some additional functions in v5 might help. As always, thanks in anticipation.

[cooldude832]

Pulling out my analytical chemistry book excel's "slope" equation is said to calculate using

m = Matrix{ sigma(Xiyi)  sigma(Xi) // Sigma(Yi)  n}  divided by D

 

b = MATRIX{ sigma(X^2i)  sigma(XiYi) // sigma(Xi)  sigma(Yi)} divdied by D

 

D is defined as the 2x2 matrix{sigma(x(^2)i) sigma(xi) // sigma(xi)  n}

 

 

So  using the 2 arrays you need to develop the sigmas of xiyi xi, yi x^2i, yi  and get your n and then solve your matrixs and your done

 

Its a lot of plug n chug

 

 

 

PEAR might be able to calculate this m and b for you

[cooldude832]

here is a good start to it

<?php
function slope($d_set1, $d_set2){
$XY = array()
foreach($d_set1 as $key=>$value){
	$XY[$key] = $value*$d_set2[$key];
}
$X2 = array();
foreach($d_set1 as $key => $value){
	$X2[$key] = pow($value,2);
}

$Xi = array_sum($d_set1);
$Yi = array_sum($d_set2);
$XiYi = array_sum($XY);
$X2i = array_sum($X2);
}
?>

[virtual_odin]

I've found the below, which seems to be pretty much what you are suggesting.

 

Does your book give any help on R squared?  Thanks.

 

/**

* linear regression function

* @param $x array x-coords

* @param $y array y-coords

* @returns array() slope=>slope, intercept=>intercept

*/

function linear_regression($x, $y) {

 

 // calculate number points

 $n = count($x);

 

 // ensure both arrays of points are the same size

 if ($n != count($y)) {

 

   trigger_error("linear_regression(): Number of elements in coordinate arrays do not match.", E_USER_ERROR);

 

 }

 

 // calculate sums

 $x_sum = array_sum($x);

 $y_sum = array_sum($y);

 

 $xx_sum = 0;

 $xy_sum = 0;

 

 for($i = 0; $i < $n; $i++) {

 

   $xy_sum+=($x[$i]*$y[$i]);

   $xx_sum+=($x[$i]*$x[$i]);

 

 }

 

 $m = (($n * $xy_sum) - ($x_sum * $y_sum)) / (($n * $xx_sum) - ($x_sum * $x_sum));

 

 $b = ($y_sum - ($m * $x_sum)) / $n;

 

 return array("slope"=>$m, "intercept"=>$b);

 

}

 

var_dump( linear_regression(array(1, 2, 3, 4), array(1.5, 2.1, 1.3, 1.4)) );