3 Big Data Analytics - Homework 3 Question 3. [40 points] We use KNN 7 to predict the class (LEAVE/ STAY) using the dataset CHURN. (a) Warmup: Copy the IRIS code and implement the KNN algorithm. What does the last line mean? library(caret) data("iris") data = iris set.seed (1) ind = sample(seq(1,150,1), 100, replace=FALSE) xtraining = data[ind, 1:41 ytraining = data[ind, 5] xtesting = data[-ind, 1:4] ytesting = data[-ind, 5] knn = knn3(xtraining, ytraining, k=3) knn y = predict(knn, newdata=xtesting, type=c('class')) sum(as. numeric(y == ytesting) (b) Why it is not good to choose an even number as k? (c) Load the dataset churn.arff' into R. (d) Set the seed number equal to 1, and split the data equally into training and testing. The training and testing data for features contain columns 2 to 7 only. (e) Use knn3 function in caret to implement knn algorithm using k=3. Report the accuracy rate. Use knn3 function in caret to implement knn algorithm using k=5. Report the accuracy rate. (9) Use knn3 function in caret to implement knn algorithm using k=11. Report the accuracy rate. (h) How many neighbors k will you choose? You can get the 'churn.arff @https://ws28.cn/f/4yi36ulxs4o 3 Big Data Analytics - Homework 3 Question 3. [40 points] We use KNN 7 to predict the class (LEAVE/ STAY) using the dataset CHURN. (a) Warmup: Copy the IRIS code and implement the KNN algorithm. What does the last line mean? library(caret) data("iris") data = iris set.seed (1) ind = sample(seq(1,150,1), 100, replace=FALSE) xtraining = data[ind, 1:41 ytraining = data[ind, 5] xtesting = data[-ind, 1:4] ytesting = data[-ind, 5] knn = knn3(xtraining, ytraining, k=3) knn y = predict(knn, newdata=xtesting, type=c('class')) sum(as. numeric(y == ytesting) (b) Why it is not good to choose an even number as k? (c) Load the dataset churn.arff' into R. (d) Set the seed number equal to 1, and split the data equally into training and testing. The training and testing data for features contain columns 2 to 7 only. (e) Use knn3 function in caret to implement knn algorithm using k=3. Report the accuracy rate. Use knn3 function in caret to implement knn algorithm using k=5. Report the accuracy rate. (9) Use knn3 function in caret to implement knn algorithm using k=11. Report the accuracy rate. (h) How many neighbors k will you choose? You can get the 'churn.arff @https://ws28.cn/f/4yi36ulxs4o