1
I’m studying clusters using R.
My question is whether, to classify new entrants with an already performed Clusterization, the function Knn can be used, whether for hierarchical or partitional clusters.
1 answer
3
It depends on which package you are using Knn, but usually to evaluate the model you use the function predict(modelo, novo_dado).
Follow an example using the package caret who already chooses the best parameters automatically:
library(caret)
# fixar semente RNG para resultado reprodutível
set.seed(123)
# treinar modelo knn utilizando a coluna Species como saída
knn_model <- train(
Species ~ .,
data = iris,
method = 'knn'
)
print("Resultado do treinamento:")
print(knn_model)
print("Predição da primeira linha:")
print(predict(knn_model, iris[1,]))
Exit:
[1] "Resultado do treinamento:"
k-Nearest Neighbors
150 samples
4 predictor
3 classes: 'setosa', 'versicolor', 'virginica'
No pre-processing
Resampling: Bootstrapped (25 reps)
Summary of sample sizes: 150, 150, 150, 150, 150, 150, ...
Resampling results across tuning parameters:
k Accuracy Kappa
5 0.9583142 0.9364989
7 0.9573900 0.9350381
9 0.9599157 0.9391095
Accuracy was used to select the optimal model using the
largest value.
The final value used for the model was k = 9.
[1] "Predição da primeira linha:"
[1] setosa
Levels: setosa versicolor virginica
-
Lucas, thanks for your help. I’ll test it here. Hugs.
Browser other questions tagged r clusterization
You are not signed in. Login or sign up in order to post.
I don’t know anything about R, but as far as I know Knn is not used to classify/evaluate an unknown entrant.
– Victor Stafusa
Victor, thanks for the support. I understand that this function serves to classify new entrants. My question is whether I can use it for hierarchical clusters (e.g., hclust()) as well as partitional (e.g., kmeans).
– Raul Albuquerque