K- Nearest Neighbors.
K-nn algorithm is a very basic classification algorithm. It can be used for both regression and classification. It is a supervised learning algorithm. KNN is easy to interpret and fairs well across calculation time.
Intuition behind knn.
Let us take an example where we are given features of a particular type of flower (like sepal length,sepal width,petal length,petal width) and its class (like Iris Setosa ,Iris Versicolour, Iris Virginica). Our task is to classify a new flower on the basis of its attributes in one of the class.
For this we select a k value. For example k=3 which means we are selecting 3 points which are least distant from the new point. To measure the least distance we generally use Euclidean distance. the maximum no points which are closest to the new point are calculated.This determines the class of new point.
— KNN is non parametric which means it does not make any assumption as practical data.
— The biggest use of KNN is recommender system.
KNN is a lazy learner or lazy algorithm — KNN is lazy because it doesn’t learn a discriminative function from the training data but “memorizes” the training dataset instead.