BenHeubl

boruta_signif <-names(boruta_output_Boruta$finalDecision[boruta_output_Boruta$finalDecision %in% c("Confirmed", "Tentative")])
print(boruta_signif)

#Plot:
plot(boruta_output_Boruta, cex.axis=.7, las=2, xlab="", main="Variable Importance")

BenHeubl

boruta_output_Boruta <- Boruta(Percent.Leave ~ ., data=na.omit(Brexit), doTrace=2)
boruta_output_Boruta

BenHeubl

# We run this with the updated model
Brexit_variable_importance <- data.frame(imp = Brexit_model_final$variable.importance)
Brexit_variable_importance2 <- Brexit_variable_importance %>% 
  tibble::rownames_to_column() %>% 
  dplyr::rename("variable" = rowname) %>% 
  dplyr::arrange(imp) %>%
  dplyr::mutate(variable = forcats::fct_inorder(variable))
ggplot2::ggplot(Brexit_variable_importance2) +
  geom_col(aes(x = variable, y = imp),
           col = "white", fill = "navy blue", show.legend = F) +

BenHeubl

Brexit_test$pred <- predict(Brexit_model_final, Brexit_test, type = "class")
base_accuracy <- mean(Brexit_test$pred == Brexit_test$Percent.Leave)

base_accuracy

BenHeubl

Brexit_model_final <- rpart(Percent.Leave ~ . , 
data = Brexit, method = "class", 
control = rpart.control(cp = TYPE_YOUR_CP_VALUE))

BenHeubl

printcp(Brexit_model)
plotcp(Brexit_model)

BenHeubl

#We need to quickly produce a model again - a conditional inference tree. Documentation here (we will switch back to rpart shortly after this)
install.packages("ggparty")
install.packages("partykit")
library(ggparty)
library(partykit)

Brexit_model_ggparty <- partykit::ctree(Percent.Leave ~ ., data = Brexit_train, control = ctree_control(testtype = "Teststatistic"))
ggparty(Brexit_model_ggparty) +
 geom_edge() +
 geom_edge_label() +

BenHeubl

#decision tree plot:
rpart.plot(Brexit_model, type = 4, extra = 0, branch.lty = 3, box.palette = "RdYlGn")

BenHeubl

#load packages:
install.packages('rpart')
install.packages('rpart.plot')
library(rpart)
library(rpart.plot)

#This is our base model:
Brexit_model <- rpart(Percent.Leave ~ . , data = Brexit_train, method = "class", control = rpart.control(cp = 0))
Brexit_model

BenHeubl

#Test and training dataset
sample_ind <- sample(nrow(Brexit), (nrow(Brexit) * 0.70))

Brexit_train <- Brexit[sample_ind,]
Brexit_test <- Brexit[-sample_ind,]