K-Nearest Neighbors

Introduction

K-Nearest Neighbors (KNN) classifies points based on the majority class of their k nearest neighbors.

Building KNN

library(caret)

# Train model
knn_model <- train(target ~ ., data = train, method = "knn")

# Different k values
knn_model <- train(target ~ ., data = train, 
                   method = "knn",
                   tuneGrid = data.frame(k = 1:10))

# Preprocessing
train(target ~ ., data = train, method = "knn",
      preProcess = c("center", "scale"))

Predictions

pred <- predict(knn_model, test)

Choosing K

# Cross-validation
train(target ~ ., data = train, method = "knn",
      trControl = trainControl(method = "cv"))

Summary

KNN is simple and effective. Choose k based on cross-validation.