How to input Pre-Clustered data for K-Nearest Neighbor algorithm

Question

I have a question about a project that I am working on. I am new to RI will appreciate any help I can get.

I am applying k-nearest neighbor algorithm on the data(using knn function from the class library). I have read somewhere that you can increase the efficiency of k-nearest neighbor classification by pre-clustering the data. My question is that if I use k-means to cluster the data, how do I input the data to k-nearest neighbor algorithm? Do I input the result of kmeans function ie the dataset with cluster means or do I combine the assigned cluster to the data and then run k-nearest neighbor on each cluster separately? Please give some guidance.

I am new to the site, so I am not sure how to properly ask the question. The code for K-Nearest Neighbor is below. The data has a lot of variables, so I am not putting that.

I am trying to predict the Lung Cancer Malignancy for easy and hard cases. The ClusterSIM (labels) has 2 classes 0 (Easy) and 1 (Hard).

library(caret)
#Doing stratified Random Sampling
inTrain = createDataPartition(LIDC$ClusterSIM, p = 6.6/10, list = FALSE)

train1=LIDC[inTrain,]
train = train1[-65] #Removing the CLusterSIM (labels)column from training data
cc = train1[65] #creating a data.frame for ClusterSIM
c1 = as.vector(cc$ClusterSIM) # convert CLusterSIM (Labels) to a vector

test1 =LIDC[-inTrain,] #creating testing data
test = test1[-65] #removing labels from testing data

knn_predict = knn(train, test, c1, k=25)

#converting labels for testing data into factors so I can create a table.
test_val = test1[65]
test_val_factor = as.factor(test_val$ClusterSIM)

table(test_val_factor, knn_predict)

Answer 1

There are several libraries for knn, class is just one. [*]

Looking at the doc for class::knn , it doesn't look like you can give it preclusters.

So search for a different library.

[*] in fact, in R there are often several competing and incompatible libraries for doing anything.