[R] XGBoost.ipynb

Kernel: R

XGBoost

Data preprocessing

# Importing the dataset
dataset = read.csv('Churn_Modelling.csv')

head(dataset, 5)

dataset = dataset[4:14]
dataset = dataset[-c(3)]

head(dataset, 5)

# Encoding the categorical variables as factors
dataset$Geography = as.numeric(factor(dataset$Geography,
                                      levels = c('France', 'Spain', 'Germany'),
                                      labels = c(1, 2, 3)))

# Splitting the dataset into the Training set and Test set
# install.packages('caTools')
library(caTools)
set.seed(1234)
split = sample.split(dataset$Exited, SplitRatio = 0.8)
training_set = subset(dataset, split == TRUE)
test_set = subset(dataset, split == FALSE)

Fitting XGBoost to the Training set

# install.packages('xgboost')
library(xgboost)

classifier = xgboost(data = as.matrix(training_set[-10]), 
                     label = training_set$Exited,
                     nrounds = 10)

[1]	train-rmse:0.417976 
[2]	train-rmse:0.369643 
[3]	train-rmse:0.341009 
[4]	train-rmse:0.325538 
[5]	train-rmse:0.316370 
[6]	train-rmse:0.309533 
[7]	train-rmse:0.306012 
[8]	train-rmse:0.302703 
[9]	train-rmse:0.300868 
[10]	train-rmse:0.298456 

---

Its better to think that the converging point is 0.30. But it is worth noting that we can decrease the train-rmse(which stands for root mean square deviation) if we keep on inscreasing nrounds. I tried it myself but the result will be too long to display here. Better try it by running this (below code is just for cross-validation to assess model performance):

set.seed(1234)
cv = xgb.cv(data = as.matrix(training_set[-10]), 
             label = training_set$Exited, 
             nfold = 5,
             nrounds = 500)

Also if you don't want to display train-rmse then pass verbose = 0 as a parameter.

---

Applying k-Fold Cross Validation

library(caret)

Loading required package: lattice
Loading required package: ggplot2

folds = createFolds(training_set$Exited, k = 10)
cv = lapply(folds, function(x){
    training_fold = training_set[-x, ]
    test_fold = training_set[x, ]
    classifier = xgboost(data = as.matrix(training_set[-10]), 
                         label = training_set$Exited,
                         nrounds = 10)
    y_fold_pred = predict(classifier, newdata = as.matrix(test_fold[-10]))
    y_fold_pred = (y_fold_pred > 0.5)
    cm = table(test_fold[, 10], y_fold_pred)
    accuracy = (cm[1, 1] + cm[2, 2])/(sum(cm))
    return(accuracy)
})

[1]	train-rmse:0.417976 
[2]	train-rmse:0.369643 
[3]	train-rmse:0.341009 
[4]	train-rmse:0.325538 
[5]	train-rmse:0.316370 
[6]	train-rmse:0.309533 
[7]	train-rmse:0.306012 
[8]	train-rmse:0.302703 
[9]	train-rmse:0.300868 
[10]	train-rmse:0.298456 
[1]	train-rmse:0.417976 
[2]	train-rmse:0.369643 
[3]	train-rmse:0.341009 
[4]	train-rmse:0.325538 
[5]	train-rmse:0.316370 
[6]	train-rmse:0.309533 
[7]	train-rmse:0.306012 
[8]	train-rmse:0.302703 
[9]	train-rmse:0.300868 
[10]	train-rmse:0.298456 
[1]	train-rmse:0.417976 
[2]	train-rmse:0.369643 
[3]	train-rmse:0.341009 
[4]	train-rmse:0.325538 
[5]	train-rmse:0.316370 
[6]	train-rmse:0.309533 
[7]	train-rmse:0.306012 
[8]	train-rmse:0.302703 
[9]	train-rmse:0.300868 
[10]	train-rmse:0.298456 
[1]	train-rmse:0.417976 
[2]	train-rmse:0.369643 
[3]	train-rmse:0.341009 
[4]	train-rmse:0.325538 
[5]	train-rmse:0.316370 
[6]	train-rmse:0.309533 
[7]	train-rmse:0.306012 
[8]	train-rmse:0.302703 
[9]	train-rmse:0.300868 
[10]	train-rmse:0.298456 
[1]	train-rmse:0.417976 
[2]	train-rmse:0.369643 
[3]	train-rmse:0.341009 
[4]	train-rmse:0.325538 
[5]	train-rmse:0.316370 
[6]	train-rmse:0.309533 
[7]	train-rmse:0.306012 
[8]	train-rmse:0.302703 
[9]	train-rmse:0.300868 
[10]	train-rmse:0.298456 
[1]	train-rmse:0.417976 
[2]	train-rmse:0.369643 
[3]	train-rmse:0.341009 
[4]	train-rmse:0.325538 
[5]	train-rmse:0.316370 
[6]	train-rmse:0.309533 
[7]	train-rmse:0.306012 
[8]	train-rmse:0.302703 
[9]	train-rmse:0.300868 
[10]	train-rmse:0.298456 
[1]	train-rmse:0.417976 
[2]	train-rmse:0.369643 
[3]	train-rmse:0.341009 
[4]	train-rmse:0.325538 
[5]	train-rmse:0.316370 
[6]	train-rmse:0.309533 
[7]	train-rmse:0.306012 
[8]	train-rmse:0.302703 
[9]	train-rmse:0.300868 
[10]	train-rmse:0.298456 
[1]	train-rmse:0.417976 
[2]	train-rmse:0.369643 
[3]	train-rmse:0.341009 
[4]	train-rmse:0.325538 
[5]	train-rmse:0.316370 
[6]	train-rmse:0.309533 
[7]	train-rmse:0.306012 
[8]	train-rmse:0.302703 
[9]	train-rmse:0.300868 
[10]	train-rmse:0.298456 
[1]	train-rmse:0.417976 
[2]	train-rmse:0.369643 
[3]	train-rmse:0.341009 
[4]	train-rmse:0.325538 
[5]	train-rmse:0.316370 
[6]	train-rmse:0.309533 
[7]	train-rmse:0.306012 
[8]	train-rmse:0.302703 
[9]	train-rmse:0.300868 
[10]	train-rmse:0.298456 
[1]	train-rmse:0.417976 
[2]	train-rmse:0.369643 
[3]	train-rmse:0.341009 
[4]	train-rmse:0.325538 
[5]	train-rmse:0.316370 
[6]	train-rmse:0.309533 
[7]	train-rmse:0.306012 
[8]	train-rmse:0.302703 
[9]	train-rmse:0.300868 
[10]	train-rmse:0.298456 

mean(as.numeric(cv)) # accuracy

sd(as.numeric(cv)) # Standard deviation