Created
March 2, 2024 07:02
-
-
Save defrindr/c03dcdbbebd62ae0d35f602735695ecf to your computer and use it in GitHub Desktop.
DecisionTreeClassifier
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
<?php | |
namespace App\Helpers; | |
use Illuminate\Database\Eloquent\Collection; | |
class DecisionTreeClassifier | |
{ | |
protected Collection $dataFrame; | |
protected array $predict; | |
public function fit($dataFrame) | |
{ | |
$this->dataFrame = $dataFrame; | |
} | |
/** | |
* Menghitung entropi dari sebuah kolom | |
*/ | |
function calculate_entropy($probabilities) | |
{ | |
$entropy = 0; | |
foreach ($probabilities as $probability) { | |
if ($probability != 0) { | |
$entropy += $probability * log($probability, 2); | |
} | |
} | |
return -$entropy; | |
} | |
function calculate_gain_information($parent_entropy, $child_entropies, $child_probabilities) | |
{ | |
$gain = $parent_entropy; | |
foreach ($child_entropies as $i => $child_entropy) { | |
$gain -= $child_probabilities[$i] * $child_entropy; | |
} | |
return $gain; | |
} | |
/** | |
* Mencari probabilitas dari sebuah kolom | |
*/ | |
function find_probabilities($feature) | |
{ | |
$probabilities = []; | |
foreach ($this->dataFrame as $data) { | |
$keyword = $data[$feature]; | |
if (!isset($probabilities[$keyword])) { | |
$probabilities[$keyword] = 1; | |
} else { | |
$probabilities[$keyword]++; | |
} | |
} | |
$totalData = $this->totalDataFrame(); | |
foreach ($probabilities as $index => $count) { | |
$probabilities[$index] = $count / $totalData; | |
} | |
return $probabilities; | |
} | |
public function totalDataFrame() | |
{ | |
return $this->dataFrame->count(); | |
} | |
public function predict(array $filter) | |
{ | |
$features = array_keys($filter); | |
$featureParentEntropies = []; | |
$entropies = []; | |
$probabilities = []; | |
foreach ($features as $feature) { | |
$probability = $this->find_probabilities($feature); | |
$probabilities[$feature] = $probability; | |
$parentEntropy = $this->calculate_entropy($probability); | |
$featureParentEntropies[$feature] = $parentEntropy; | |
foreach ($probability as $item => $value) { | |
$entropies[$feature][$item] = $this->calculate_entropy([$value, 1 - $value]); | |
} | |
} | |
foreach ($features as $feature) { | |
$childEntropy = 0; | |
$childProbability = 0; | |
$childEntropy = $entropies[$feature]; | |
$childProbability = $probabilities[$feature]; | |
$gainInformation[$feature] = $this->calculate_gain_information($featureParentEntropies[$feature], $childEntropy, $childProbability); | |
} | |
// order descending | |
arsort($gainInformation); | |
$dataFrame = $this->dataFrame; | |
foreach (array_keys($gainInformation) as $column) { | |
$dataFrame = $dataFrame->filter(function ($data) use ($filter, $column) { | |
if (is_array($filter[$column])) { | |
return in_array($data[$column], $filter[$column]); | |
} | |
return $data->$column == $filter[$column]; | |
}); | |
} | |
return $dataFrame; | |
} | |
} |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment