/**
* Calculate probability of feature in category.
*
* @param string $category category
* @param string $feature feature
*
* @return float
*/
protected function featureProbability($category, $feature)
{
if ($this->storage->getCategoryFeaturesCount($category) == 0) {
// avoid `0` probability
return 0.01;
}
// P(category|feature) = P(feature|category) * P(category) /
// P(feature|category) * P(category) +
// ( P(feature|otherCategories) * P(otherCategories))
// P(feature|category)
// = feature frequency in category / all features in category
$featureCountInCategory = $this->storage->getFeatureCountInCategory($category, $feature);
$categoryFeaturesCount = $this->storage->getCategoryFeaturesCount($category);
// P(feature|category)
$featureCategoryProbability = $this->storage->getFeatureDocumentsCountInCategory($category, $feature) / $this->storage->getCategoryDocumentsCount($category);
if ($featureCategoryProbability == 0) {
$featureCategoryProbability = 0.01;
}
// P(category)
$categoryProbability = $this->storage->getCategoryDocumentsCount($category) / $this->storage->getDocumentsCount();
// P(otherCategories)
$categoryImprobability = ($this->storage->getFeatureCount($feature) - $featureCountInCategory) / ($this->storage->getFeaturesCount() - $categoryFeaturesCount);
$categoryImprobability = $categoryImprobability == 0 ? 0.01 : $categoryImprobability;
// P(category|feature)
$probability = $featureCategoryProbability * $categoryProbability / ($featureCategoryProbability * $categoryProbability + $categoryImprobability * (1 - $categoryProbability));
return $probability;
}
public function testDocumentsCount()
{
for ($i = 1; $i < 4; $i++) {
$this->storage->increaseDocumentsCount();
$this->assertEquals($i, $this->storage->getDocumentsCount());
}
}
