k-NN (EN)

k-NN Algorithm

The k-Nearest Neighbor (k-NN) algorithm is a simple but effective learning algorithm for classification and regression tasks. It classifies data points based on the classes of the k nearest neighbors in the feature space. The number of neighbors (k) is an important parameter that influences the complexity and accuracy of the model. k-NN is a so-called 'lazy learner' as it does not perform an explicit training phase, but rather performs all calculations at prediction time.

Architecture

flowchart TD   A[Data point] --> B[Distance calculation]   B --> C{Find k-neighbors}   C --> D[Classification/Regression]   D --> E[Prediction] 

In Context

• Typically used together with metrics such as Euclidean distance or Manhattan distance
• Related to: Lazy Learning, Instance-Based Learning, Feature Engineering
• Example use: Image recognition, recommendation systems, anomaly detection