Let me give some background context: I have a list of 100 vectors, each one with 50 dimensions, and I'd like to associate the first 50 vectors with class A and the last ones with class B.
My question is: How can I do it in order to apply kNN later and which library has a kNN method more appropriate for this?
Thanks in advance.
One of the first options that come to my mind is to make a data.frame from a list of vectors, create factor indicator and then use knn from class package.
Make a data.frame from a list of vectors
Using rbind, make a matrix and then use as.data.frame function (more examples in this question). Assuming that l is a list of vectors:
data <- as.data.frame(do.call(rbind, l))
Factor indicator
class <- as.factor(c(rep("A", 50), rep("B", 50)))
kNN classification using class package
In case you don't have separate data for testing, probably the best way would be
train.ind <- sample(1:100, 75) # making indexes to split data into 75% train and 25% test
resulting.classes <- knn(train = data[train.ind, ], test = data[-train.ind, ], cl = class)
And if you have separate train and test data then just use
resulting.classes <- knn(train = train.data, test = test.data, cl = class)
Other choices for kNN that might be useful - ‘kknn’ package and 'FNN' package., but class package seems the easiest one for simple kNN classification.
