Machine LearningLesson 91Free preview

K-nearest neighbors

This lesson is part of Python · All levels

Summary

K-Nearest Neighbors (KNN) classifies new data based on the majority of k's nearest neighbors in the feature space. When the prediction is made, the model calculates the distance of the new data to the entire training data, sorts it, and then takes the nearest k to select the class. The k-value that is too small makes the model sensitive to noise, while too large makes the boundary of the cluster blurry. KNN is a lazy algorithm because it doesn't train anything, all the calculations happen when the prediction is made.

Key points

KNN determines the class of the majority of k's nearest neighbors
The k-value sets the balance between sensitivity and noise sensitivity
The feature is mandatory discaling because the KNN depends on the distance
KNN doesn't train models, calculations are done when prediction is in order.

Full lesson

Code examples and quizzes

Full lessons, code examples, and quizzes are available to subscribers. 5 questions.