[ADP] K-NN (K-Nearest Neighbors)

> # 15. K-NN (K-Nearest Neighbors)
> 
> 
> # KNN은 데이터의 지역 구조(Local Structure)에 민감하다.
> 
> # iris dataset의 KNN
> library(DMwR)
> 
> df <- sample(1:nrow(iris), as.integer(0.7*nrow(iris)))
> trainiris <- iris[df,]
> testiris <- iris[-df,]
> 
> nc3 <- kNN(Species~., trainiris, testiris, stand = FALSE, k=3) # k=3일 때, stand는 정규화 옵션
> table(testiris[,'Species'], nc3)
            nc3
             setosa versicolor virginica
  setosa         18          0         0
  versicolor      0         13         0
  virginica       0          1        13
> 
> nc5 <- kNN(Species~., trainiris, testiris, stand = FALSE, k=5) # k=5일 때
> table(testiris[,'Species'], nc5)
            nc5
             setosa versicolor virginica
  setosa         18          0         0
  versicolor      0         13         0
  virginica       0          1        13

 

 

 

 

 

출처 : 2020 데이터 분석 전문가 ADP 필기 한 권으로 끝내기