application

source("https://bioconductor.org/biocLite.R")
biocLite("graph")
biocLite("RBGL")
biocLite('Rgraphviz')

install.packages('gRain')

library("Rgraphviz")
library('gRain')

library(bnlearn)

d <- read.csv('/home/max/Documents/M1/Bayesian Networks/A2/student-por.csv', sep=';')
d$G1 <- NULL
d$G2 <- NULL
d$Dalc <- NULL
d$Walc <- NULL
d$sex <- NULL
d$guardian <- NULL
d$nursery <- NULL
d$internet <- NULL
d$address <- NULL
d$reason <- NULL
d$famsize <- NULL
d$Pstatus <- NULL

library(Hmisc)
d$age <- cut2(d$age, g = 4, oneval=FALSE)
d$absences <- cut2(d$absences, g = 4, oneval=FALSE)
d$G3 <- cut2(d$G3, g = 4, oneval=FALSE)

d$Medu <- as.factor(d$Medu)
d$Fedu <- as.factor(d$Fedu)
d$traveltime <- as.factor(d$traveltime)
d$studytime <- as.factor(d$studytime)
d$failures <- as.factor(d$failures)
d$famrel <- as.factor(d$famrel)
d$freetime <- as.factor(d$freetime)
d$goout <- as.factor(d$goout)
d$health <- as.factor(d$health)

for( i in colnames(d) ){
	print(i)
	print(table(d[,i]))
}

V <- c("absences","activities","age","failures","famrel","famsup","Fedu","Fjob","freetime","G3","goout","health","higher","Medu","Mjob","paid","romantic","school","schoolsup","studytime","traveltime")
sp.dag = new("graphNEL", nodes = V, edgemode = "directed")

sp.dag = addEdge("failures", "G3", sp.dag)
sp.dag = addEdge("higher", "G3", sp.dag)
sp.dag = addEdge("Medu", "Fjob", sp.dag)
sp.dag = addEdge("Mjob", "Fjob", sp.dag)
sp.dag = addEdge("Fedu", "Fjob", sp.dag)
sp.dag = addEdge("Fedu", "Mjob", sp.dag)
sp.dag = addEdge("Medu", "Mjob", sp.dag)
sp.dag = addEdge("Fedu", "Medu", sp.dag)
sp.dag = addEdge("Fjob", "famrel", sp.dag)
sp.dag = addEdge("goout", "freetime", sp.dag)
sp.dag = addEdge("higher", "failures", sp.dag)
sp.dag = addEdge("higher", "school", sp.dag)
sp.dag = addEdge("higher", "studytime", sp.dag)
sp.dag = addEdge("school", "failures", sp.dag)
sp.dag = addEdge("schoolsup", "school", sp.dag)
sp.dag = addEdge("school", "traveltime", sp.dag)
sp.dag = addEdge("Fedu", "higher", sp.dag)
sp.dag = addEdge("Medu", "higher", sp.dag)

plot(sp.dag)

sp.gr <- compile(grain(sp.dag, d))

summary(sp.gr)

predict(sp.gr, response = c("G3"), newdata = d, predictors = c("absences","activities","age","failures","famrel","famsup","Fedu","Fjob","freetime","goout","health","higher","Medu","Mjob","paid","romantic","school","schoolsup","studytime","traveltime")
, type = "class")