No, it doesn't and therefore is no viable option for many real-world datasets.