5 K-nearest Neighbours Classification

5.1 Introduction

The K-nearest Neighbours (KNN) for classification, uses a similar idea to the KNN regression. For KNN, a unit will be classified as the majority of its neighbours.

5.2 Readings

Read the following chapters of An introduction to statistical learning:

5.3 Practical session

TASK - KNN classification

With the bmd.csv dataset, we will use KNN (k=3) with the variables AGE, SEX, BMI and BMD to classify FRACTURE and compute the confusion matrix

First, let’s import the dataset

#libraries that we will need
library(class)   #knn 
set.seed(1974) #fix the random generator seed 

#read the data
bmd.data     <- 
  read.csv("https://www.dropbox.com/s/c6mhgatkotuze8o/bmd.csv?dl=1", 
           stringsAsFactors = TRUE)
  
bmd.data$bmi <- bmd.data$weight_kg / (bmd.data$height_cm/100)^2

let’s standardise all the variables so they have mean 0 and SD=1 so that the distances are in the same scale.

  bmd.data$age.std     <- scale(bmd.data$age)
  bmd.data$sex.num.std <- scale(as.numeric(bmd.data$sex)) #1-F, #2-M
  bmd.data$bmi.std     <- scale(bmd.data$bmi)  
  bmd.data$bmd.std     <- scale(bmd.data$bmd)

Now we use knn=3

model.knn3 <-  knn(train = bmd.data[c("age.std", "sex.num.std",
                                      "bmi.std","bmd.std")],
                   test  = bmd.data[c("age.std", "sex.num.std",
                                      "bmi.std","bmd.std")],
                    cl    = bmd.data$fracture, k=3)
table(model.knn3, bmd.data$fracture)
##              
## model.knn3    fracture no fracture
##   fracture          38           7
##   no fracture       12         112

TRY IT YOURSELF:

  1. Repeat the classification model from above but now with k=20 and compute the confusion matrix.
See the solution code 
  model.knn20 <-  knn(train = bmd.data[c("age.std", "sex.num.std",
                                        "bmi.std","bmd.std")],
                   test  = bmd.data[c("age.std", "sex.num.std",
                                         "bmi.std","bmd.std")],
                    cl    = bmd.data$fracture, k=20 )
  table(model.knn20, bmd.data$fracture)

  1. (hard) Plot the error rate for KNN with k=1 to 20 using the variables AGE, SEX, BMI and BMD to classify FRACTURE
See the solution code 
knn.fit <- function(k.par){
  knn.model <- knn(train = bmd.data[c("age.std", "sex.num.std",
                                      "bmi.std","bmd.std")],
                   test  = bmd.data[c("age.std", "sex.num.std",
                                      "bmi.std","bmd.std")],
                    cl    = bmd.data$fracture, k=k.par )
  class.error<- 1-sum(diag(table( knn.model, bmd.data$fracture))/169)
  return(class.error)
}

  class.error1to20 <- sapply(seq(1,20), knn.fit)
  plot(seq(1,20), class.error1to20, type="l")

5.4 Exercises

  1. The dataset bdiag.csv, included several imaging details from patients that had a biopsy to test for breast cancer.
    The variable diagnosis classifies the biopsied tissue as M = malignant or B = benign.

    1. Use a KNN with k=5 to predict Diagnosis using texture_mean and radius_mean.

    2. Build the confusion matrix for the classification above

    3. Plot the scatter plot for texture_mean and radius_mean and draw the border line for the prediction of Diagnosis based on the model in a)

    4. Plot the scatter plot for texture_mean and radius_mean and draw the border line for the prediction of Diagnosis based knn, k=15