# Predmet: "Analyza zhlukov a klasifikacia dat"
# Studijny program: "Pravdepodobnost a matematicka statistika"
# Vyucujuci: Radoslav Harman, KAMS FMFI UK Bratislava
#
# Porovnanie viacerych klasifikacnych metod
# Based on "Lab 4" from the book James, G. et al.: An Introduction to Statistical Learning, Springer 2013
# See http://www-bcf.usc.edu/~gareth/ISL/ and http://www-bcf.usc.edu/~gareth/ISL/Chapter%204%20Lab.txt

library(ISLR); Smarket[1:10,]; dim(Smarket)
pairs(Smarket[,c(-4,-5,-6)],pch=19,cex=0.1)
cor(Smarket[,-9])

# Logistic Regression

attach(Smarket); help(glm)
glm.fit<-glm(Direction~Lag1+Lag2+Lag3+Lag4+Lag5+Volume,data=Smarket,family=binomial)
summary(glm.fit)
coef(glm.fit); glm.fit$coef;summary(glm.fit)$coef

glm.probs<-predict(glm.fit,type="response"); plot(glm.probs,pch=19)
glm.pred<-rep("Down",1250); glm.pred[glm.probs>.5]<-"Up"
table(glm.pred,Direction); mean(glm.pred==Direction)

train<-(Year<2005); Smarket.2005<-Smarket[!train,]; dim(Smarket.2005)
glm.fit<-glm(Direction~Lag1+Lag2+Lag3+Lag4+Lag5+Volume,data=Smarket,family=binomial,subset=train)
summary(glm.fit)

glm.probs<-predict(glm.fit,newdata=Smarket.2005,type="response")
glm.pred<-rep("Down",252); glm.pred[glm.probs>.5]<-"Up"
Direction.2005<-Direction[!train]
table(glm.pred,Direction.2005);mean(glm.pred==Direction.2005)

glm.fit<-glm(Direction~Lag1+Lag2,data=Smarket,family=binomial,subset=train)
glm.probs<-predict(glm.fit,Smarket.2005,type="response")
glm.pred<-rep("Down",252); glm.pred[glm.probs>.5]<-"Up"
table(glm.pred,Direction.2005);mean(glm.pred==Direction.2005)

glm.pred<-as.factor(glm.pred)
plot(Lag1[!train],Lag2[!train],col=as.numeric(Direction[!train]))
plot(Lag1[!train],Lag2[!train],col=as.numeric(Direction[!train]),pch=19+2*abs(as.numeric(glm.pred)-as.numeric(Direction[!train])))


# Linear Discriminant Analysis

library(MASS)
lda.fit<-lda(Direction~Lag1+Lag2,data=Smarket,subset=train)
lda.fit; names(lda.fit)
mean(lda.fit$means%*%lda.fit$scaling)

lda.pred<-predict(lda.fit, Smarket.2005); names(lda.pred)
table(lda.pred$class,Direction.2005); mean(lda.pred$class==Direction.2005)
plot(lda.pred$posterior[,1])


# Quadratic Discriminant Analysis

library(MASS); help(qda)
qda.fit<-qda(Direction~Lag1+Lag2,data=Smarket,subset=train)
qda.fit; names(qda.fit)

qda.pred<-predict(qda.fit,Smarket.2005); names(qda.pred)
table(qda.pred$class,Direction.2005); mean(qda.pred$class==Direction.2005)
plot(qda.pred$posterior[,1])

plot(Lag1[!train],Lag2[!train],col=as.numeric(Direction[!train]))
plot(Lag1[!train],Lag2[!train],col=as.numeric(Direction[!train]),pch=19+2*abs(as.numeric(qda.pred$class)-as.numeric(Direction[!train])))


# K-Nearest Neighbors

library(class); help(knn)

train.X<-cbind(Lag1,Lag2)[train,]
test.X<-cbind(Lag1,Lag2)[!train,]
train.Direction<-Direction[train]

knn.pred<-knn(train.X,test.X,train.Direction,k=1)
table(knn.pred,Direction.2005)
mean(knn.pred==Direction.2005)

knn.pred<-knn(train.X,test.X,train.Direction,k=3)
table(knn.pred,Direction.2005)
mean(knn.pred==Direction.2005)

plot(Lag1[!train],Lag2[!train],col=as.numeric(Direction[!train]))
plot(Lag1[!train],Lag2[!train],col=as.numeric(Direction[!train]),pch=19+2*abs(as.numeric(knn.pred)-as.numeric(Direction[!train])))

test.alot<-cbind(runif(10000,min=-2.5,max=2.5),runif(10000,min=-2.5,max=2.5))
knn.pred<-knn(train.X,test.alot,train.Direction,k=3)
plot(test.alot[,1],test.alot[,2],col=as.numeric(knn.pred),pch=19)
knn.pred<-knn(train.X,test.alot,train.Direction,k=101)
plot(test.alot[,1],test.alot[,2],col=as.numeric(knn.pred),pch=19)