luminsports/linear-regression

Linear regression

1.2.0 2022-08-19 06:05 UTC

Requires

• php: ^8.1
• ext-bcmath: *
• brick/math: ^0.10.2

Requires (Dev)

• friendsofphp/php-cs-fixer: ^3.2.1
• phpunit/php-code-coverage: ^9.2.7
• phpunit/phpunit: ^9.0

MIT

• Dave Barnwell
• Ryan Brereton

README

A Linear regression class that uses the least squares method to approximate a straight line to a data set. Forked from davebarnwell/ml-regression-least-squares, this version uses new PHP 8.1 syntax, and BCMath for high precision.

composer install luminsports/linear-regression

Usage:

$x = [...]; // target values
$y = [...]; // observation values

$linearRegression = new \LuminSports\LinearRegression\LeastSquares($x, $y);

$slope = $linearRegression->getSlope();
$yIntercept = $linearRegression->getIntercept();
    
// return array of differences of y values from the regression line
$differences = $linearRegression->getDifferencesFromRegressionLine();

// return array of cumulative sum of the differences of y values from the regression line
$cumulativeSum = $linearRegression->getCumulativeSumOfDifferencesFromRegressionLine();

// return array of Point objects giving the x,y values of the regression line
// for current data
$regressionLine = $linearRegression->getRegressionLinePoints();

$regressionLine[0]->getX();
$regressionLine[0]->getY();

$predictedX = $linearRegression->predictX($anObservationValue);

$predictedY = $linearRegression->predictY($aTargetValue);

$rSquared = $linearRegression->getRSquared(); // Regression fit; 1 = perfect fit 0 = no fit