require(MASS)

dat = read.table(stdin(),header=T)
x1 x2 class
4 2 A
2 4 A
2 3 A
3 6 A
4 4 A
9 10 B
6 8 B
9 5 B
8 7 B
10 8 B
 

plot(dat$x1,dat$x2,type="n"); text(dat$x1,dat$x2,dat$class)

lda.mod = lda(class ~ ., data=dat, prior = c(1,1)/2)
lda.mod

lda.pred <- predict(lda.mod, dat)
lda.pred
round(lda.pred$posterior,5)
plot(lda.mod) #plots LDA_scores (lda.pred$x) 

dat2 = data.frame(cbind( x1=c(4,5,6), x2=c(6,7,7) )) 
dat2
plot( dat$x1, dat$x2, type="n"); text(dat$x1,dat$x2,dat$class)
text(dat2$x1,dat2$x2, "X", col=2)
lda.pred2 <- predict(lda.mod, newdata=dat2) #NB: varnames must match
lda.pred2


# LDA on the Iris Data
 data(iris)
 head(iris)
 summary(iris)
 by(iris,iris$Species,summary)

 lda.mod <- lda(Species ~ ., data = iris, prior = c(1,1,1)/3)
 attributes(lda.mod)
 head(lda.mod)
 lda.pred <- predict(lda.mod, iris)
 lda.pred.class <- lda.pred$class
 table(lda.pred.class)
 table(lda.pred.class,iris$Species) #"Confusion" matrix
 lda.detail <- cbind(as.character(lda.pred$class),round(lda.pred$posterior,6))
 lda.detail[1:10,] 
 lda.detail[130:140,] 
 ldahist(lda.pred$x[,1],g=iris$Species)

 plot(lda.mod)
 plot(lda.pred$x[,1], lda.pred$x[,2], col=iris$Species) #as.numeric for colors


 set.seed(1)
 train <- sample(1:dim(iris)[1], 75)
 mod <- lda(Species ~ ., iris, prior = c(1,1,1)/3, subset = train) # training model
 lda.pred = predict(object = mod,  newdata = iris[-train, ]) # predictions
 table(lda.pred$class, iris$Species[-train]) #"Confusion" matrix

 plot(mod)
 plot(lda.pred$x[,1], lda.pred$x[,2], col=iris$Species[-train]) #as.numeric for colors