[bit.ly/k-NN] The k-NN algorithm operates as follows. For a new test instance, we first compute its distance to all the N training instances, and keep a small number k of nearest neighbours. For classification, we then predict the most dominant class among the k neighbours. For regression, we compute the mean target value across the k neighbours.
Негізгі бет k-NN 2: classification and regression
Пікірлер: 8