Decision Trees using R – Bank Loan Default Prediction

The decision tree is a key challenge in R and the strength of the tree is they are easy to understand and read when compared with other models. They are being popularly used in data science problems. These are the tool produces the hierarchy of decisions implemented in statistical analysis. Statistical knowledge is required to understand the logical interpretations of the Decision tree. As we have seen the decision tree is easy to understand and the results are efficient when it has fewer class labels and the other downside part of them is when there are more class labels calculations become complexed. This course makes one become proficient to build predictive and tree-based learning models.

Decision Tree in R is a machine-learning algorithm that can be a classification or regression tree analysis. The decision tree can be represented by graphical representation as a tree with leaves and branches structure. The leaves are generally the data points and branches are the condition to make decisions for the class of data set. Decision trees in R are considered as supervised Machine learning models as possible outcomes of the decision points are well defined for the data set. It is also known as the CART model or Classification and Regression Trees. There is a popular R package known as rpart which is used to create the decision trees in R.

To work with a Decision tree in R or in layman terms it is necessary to work with big data sets and direct usage of built-in R packages makes the work easier. A decision tree is non- linear assumption model that uses a tree structure to classify the relationships. The Decision tree in R uses two types of variables: categorical variable (Yes or No) and continuous variables. The terminologies of the Decision Tree consisting of the root node (forms a class label), decision nodes(sub-nodes), terminal node (do not split further). The unique concept behind this machine learning approach is they classify the given data into classes that form yes or no flow (if-else approach) and represents the results in a tree structure. The algorithm used in the Decision Tree in R is the Gini Index, information gain, Entropy. There are different packages available to build a decision tree in R: rpart (recursive), party, random Forest, CART (classification and regression). It is quite easy to implement a Decision Tree in R.

For clear analysis, the tree is divided into groups: a training set and a test set. The following implementation uses a car dataset. This data set contains 1727 obs and 9 variables, with which classification tree is built. In this article lets tree a ‘party ‘package. The function creates () gives conditional trees with the plot function.