create a knn model in r

K-Nearest Neighbors (KNN) is a non-parametric algorithm for classification and regression. In R, we can create a KNN model using the caret package. Here's an example:

main.r
# Load the required packages
library(caret)
library(mlbench)

# Load the dataset
data(Sonar)

# Split the dataset into training and testing sets
set.seed(123)
trainIndex <- setdiff(1:nrow(Sonar), sample(1:nrow(Sonar), 100))
training <- Sonar[trainIndex, ]
testing <- Sonar[-trainIndex, ]

# Create a KNN model
knn_model <- train(Class ~ ., data = training, method = "knn",
                   preProcess = c("center","scale"),
                   trControl = trainControl(method = "cv"))

# Predict the classes of the testing set using the KNN model
predictions <- predict(knn_model, testing)

# Calculate the accuracy of the KNN model
accuracy <- confusionMatrix(predictions, testing$Class)$overall['Accuracy']
cat(sprintf("Accuracy: %.2f%%\n", accuracy * 100))
761 chars
25 lines

In this example, we loaded the Sonar dataset from the mlbench package and split it into a training set and a testing set. We then created a KNN model using the train() function from the caret package, which automatically performs cross-validation to tune the number of neighbors (k) used in the model. We applied centering and scaling using the preProcess argument to improve the performance of the model. Finally, we predicted the classes of the testing set using the KNN model and calculated its accuracy using confusion matrix.

This is just a basic example, and there are many parameters that can be tuned to improve the performance of the KNN model. The caret package provides many tools for performing feature selection, hyperparameter tuning, and model evaluation, which can be used to create more sophisticated KNN models.