descriptors = read.csv("ca_desc.csv")
experiment = read.csv("carboxylic_acids.csv",sep=";")
dataset = merge(experiment[,c(1,3)],descriptors[,1:1445],by.x="NSC",by.y="Name")

dataset = dataset[complete.cases(dataset),]

remove = c()
for(i in 3:1445){
  
  if(var(dataset[i])<0.1){
    print(names(dataset)[i])
    remove = append(remove,i)
  }
}

dataset = dataset[,-remove]

cor(dataset[,2:20])
cor(dataset$pKa,dataset[,-c(1,2)])

model = regsubsets(pKa~.,data = dataset[2:50])
summary(model)$rsq
summary(model)$adjr2
models = c()
for(i in 1:9){
  print(which(summary(model)$which[i,]==TRUE))
  f = paste(names(which(summary(model)$which[i,]==TRUE))[-1],collapse="+")
  print(f)
  models[[i]] = lm(paste("pKa~",f),data=dataset)
}

n = 9
model = regsubsets(pKa~.,data = dataset[-1],method="forward",nvmax=n)
summary(model)$rsq
summary(model)$adjr2
models = c()
for(i in 1:n){
  print(which(summary(model)$which[i,]==TRUE))
  f = paste(names(which(summary(model)$which[i,]==TRUE))[-1],collapse="+")
  print(f)
  models[[i]] = lm(paste("pKa~",f),data=dataset)
}

model = regsubsets(pKa~.,data = dataset[-1],method="backward",nvmax=n)
summary(model)$rsq
summary(model)$adjr2
models = c()
for(i in 1:n){
  print(which(summary(model)$which[i,]==TRUE))
  f = paste(names(which(summary(model)$which[i,]==TRUE))[-1],collapse="+")
  print(f)
  models[[i]] = lm(paste("pKa~",f),data=dataset)
}