/**
* @param $data
* @param $base_key
* @param $base_value
* @param $split_key
* @param $split_value
* @return DecisionTreeData
*/
private function makeDecisionTreeData($data, $base_key, $base_value, $split_key, $split_value)
{
$decisionTreeData = new DecisionTreeData();
$decisionTreeData->number = count($data);
$decisionTreeData->split_key = $split_key;
$decisionTreeData->split_value = $split_value;
$split_array = ArrayService::split_by_key($data, $base_key);
$decisionTreeData->match = isset($split_array[$base_value]) ? count($split_array[$base_value]) : 0;
$decisionTreeData->unmatch = $decisionTreeData->number - $decisionTreeData->match;
return $decisionTreeData;
}
/**
* @param $samples
* @param $attr
* @return int
*/
private function get_information_gain($samples, $attr)
{
$gain_reduction = 0;
$total_count = count($samples);
$possible_values_count = ArrayService::count_values($samples, $attr);
foreach ($possible_values_count as $value => $count) {
$e = $this->calculate_entropy($samples, $attr, $value);
$gain_reduction += $count * $e / $total_count;
}
$e = $this->calculate_entropy($samples);
$ret = $e - $gain_reduction;
return $ret;
}
