20-Discriminant2

# n cross validation to determine an estimate

This preview shows page 1. Sign up to view the full content.

This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: ;svd" [8] "call" > table(predict(diab.ld,diab[,1:5])\$class,diab[,6]) 123 1 26 0 0 2 5 31 3 3 1 5 73 . . . . . "N" . Cross-validation To determine an estimate of the misclassiﬁcation rate that is not biased, we use cross-validation. Usually for LDA we use leave-one out cross validation (n fold) X 1 ∪ X2 ∪ X 3 . . . ∪ X n . . . . . . conf <- function(class.predict,class){ confusion=table(class.predict,class) return(confusion) } library(class) train <- rbind(iris3[1:25,,1], iris3[1:25,,2], iris3[1:25,,3]) test <- rbind(iris3[26:50,,1], iris3[26:50,,2], iris3[26:50,,3]) cl <- factor(c(rep("s",25), rep("c",25), rep("v",25))) knn(train, test, cl, k = 3, prob=TRUE) iris.knncv2=knn.cv(train, cl, k = 2, prob = TRUE) iris.knncv4=knn.cv(train, cl, k = 4, prob = TRUE) iris.knncv8=knn.cv(train, cl, k = 8, prob = TRUE) iris.knncv12=knn.cv(train, cl, k = 12, prob = TRUE) table(cl,iris.knncv2) iris.knncv2 cl csv c 47 0 3 s 0 50 0 v 5 0 45 > table(cl,iris.knncv8) iris.knncv8 . . . . . . cl csv c 47 0 3 s 0 50 0 v 2 0 48 > table(cl,iris.knncv12) iris.knncv12 cl csv c 47 0 3 s 0 50 0 v 3 0 47 classifier <- IBk(class ~ ., data = iris, control = Weka_control(K = 1 > evaluate_Weka_classifier(classifier, numFolds = 10) === 10 Fold Cross Validation === === Summary === Correctly Classified Instances 144 96 % Incorrectly Classifie...
View Full Document

## This document was uploaded on 02/16/2014 for the course STATISTICS 3026 at Columbia.

Ask a homework question - tutors are online