R Data Analysis Cookbook, Second Edition - Second Edition

Using a convenience function to run KNN for multiple k values

Running knn for several values of k to choose the best one, involves repetitively executing similar lines of code several times. We can automate the process with the following convenience function that runs knn for multiple values of k, reports the RMS error for each, and also produces a scree plot of the RMS errors:

rdacb.knn.reg.multi <- function (trg_predictors, val_predictors, trg_target, val_target, start_k, end_k)  
{ 
  rms_errors <- vector() 
  for (k in start_k:end_k) { 
    rms_error <- rdacb.knn.reg(trg_predictors, val_predictors,  
                               trg_target, val_target, k) 
    rms_errors <- c(rms_errors, rms_error) 
  } 
  plot(rms_errors, type = "o", xlab = "k", ylab = "RMSE") 
} 

With the preceding function, ...