#########################################################################################################
### BMTRY 790: Machine Learning												###
### Spring 2023															###
###																	###
### R CODE for CART and Logic Regression tree-based methods.  									###
###																	###
#########################################################################################################

#####################################
###  SINGLE TREE METHODS FOR THE  ###
###  BREAST CANCER DATA SET       ###
#####################################

### CART ###
library(rpart)
library(rpart.plot)
library(caret)

### Regression Tree
air<-read.csv("H:\\public_html\\BMTRY790_Spring2023\\Datasets\\Air.csv")
Cair1<-rpart(log(Ozone)~., data=air)

plot(Cair1, compress=T, main="CART model 1 for Ozone")
text(Cair1, use.n=T)
rpart.plot(Cair1, compress=T, main="CART model 1 for Ozone")

###Tuning the CART model using “prune”
prune(Cair1, cp=0.05)

###Tuning CART on from end using rpart.control
Cair2<-rpart(log(Ozone)~., data=air, control=rpart.control(minsplit=6, minbucket=2, cp=0.01))
Cair3<-rpart(log(Ozone)~., data=air, control=rpart.control(minsplit=10, minbucket=5, cp=0.05))
	#Note rpart.control sets control parameters for fitting the CART model
	# minsplit sets min number observations that node must have for split to be tried
	# minbucket is min number observations in terminal node
	# cp is complexity parameter-> any split that doesn't imporve overall lack of fit by a factor cp not used


par(mfrow=c(1,3))
rpart.plot(Cair1, compress=T, main="CART model 1 for Ozone")
rpart.plot(Cair2, compress=T, main="CART model 2 for Ozone")
rpart.plot(Cair3, compress=T, main="CART model 3 for Ozone")

###Tuning using caret package (Data must be complete)
air2<-na.omit(air[,c(1:4,7)])
trair<-train(log(Ozone) ~ ., data=air2, method="rpart", trControl=trainControl(method="boot", number=1000))


### Classification Tree
ln<-read.csv("H:\\public_html\\BMTRY790_Spring2023\\Datasets\\LupusNephritis.csv")
Cmod1<-rpart(CR90~., data=ln, method="class", control=rpart.control(minsplit=6, minbucket=2, cp=0.05))
Cmod1
rpart.plot(Cmod1, compress=T, main="CART model for Treatment Response LN")

trln<-train(as.factor(CR90) ~ ., data=ln, method="rpart", trControl=trainControl(method="boot", number=1000))
trln


########################
### Logic Regression ###
########################
library(LogicReg)

BC<-read.csv("H:\\public_html\\BMTRY790_Spring2023\\Datasets\\BC_trees.csv")
BC$agecat<-ifelse(BC$age<61, 0,1)
BC$gradecat<-ifelse(BC$gradecat<3, 0, 1)


### Fitting a classification model
###Using CV to select number of leaves
anneal.params <- logreg.anneal.control(start = 2, end = -1, iter = 100000)
logreg(resp=BC$gradecat, bin=BC[,c(3:8,10)], type=1, select=3, ntrees=1, nleaves=c(2,8),
	 kfold=5, anneal.control = anneal.params)

anneal.params <- logreg.anneal.control(start = 1, end = -1, iter = 100000, update=1000)
fit<-logreg(resp=BC$gradecat, bin=BC[,c(3:8,10)], type=1, select=1, ntrees=1, nleaves=5,
            anneal.control = anneal.params)

anneal.params <- logreg.anneal.control(start = 2, end = -1, iter = 100000, update=1000)
fit<-logreg(resp=BC$gradecat, bin=BC[,c(3:8,10)], type=1, select=1, ntrees=1, nleaves=5,
            anneal.control = anneal.params)

plot(fit)

### Fitting a logistic regression tree(s) instead of a classification tree
###Using CV to select number of leaves

anneal.params <- logreg.anneal.control(start = 2, end = -4, iter = 100000)
logreg(resp=BC$gradecat, bin=BC[,c(3:8,10)], type=3, select=3, ntrees=c(1,2), nleaves=c(3,8),
	 kfold=5, anneal.control = anneal.params)

anneal.params <- logreg.anneal.control(start = 2, end = -4, iter = 100000, update=5000)
fit<-logreg(resp=BC$gradecat, bin=BC[,c(3:8,10)], type=3, select=1, ntrees=1, nleaves=7,
            anneal.control = anneal.params)

anneal.params <- logreg.anneal.control(start = 1, end = -3.5, iter = 100000, update=5000)
fit<-logreg(resp=BC$gradecat, bin=BC[,c(3:8,10)], type=3, select=1, ntrees=1, nleaves=7,
            anneal.control = anneal.params)

fit

plot(fit)