Learning groupings within data and classifying with kNN can be done using the following steps:

1. Scale the data and remove non-numeric columns:

set.seed(123)
scaled_iris <- iris %>% mutate_if( is.numeric, .funs = scale)
labels <- scaled_iris$Species
scaled_iris$Species <- NULL

2. Extract a training and test dataset:

train_rows <- sample(nrow(scaled_iris), 0.8 * nrow(scaled_iris), replace = FALSE)
train_set <- scaled_iris[train_rows, ]
test_set <- scaled_iris[-train_rows, ]
train_labels <- labels[train_rows]
test_set_labels <- labels[-train_rows]

3. Make the model and predictions on the test set:

test_set_predictions <- knn(train = train_set, test = test_set, cl = train_labels, k = 10)

4. Compare the prediction with the actual class:

caret::confusionMatrix(test_set_predictions,  test_set_labels)