Decision tree variable selection

Hi everybody I have a question regarding the variable selection in a decision three, I know before hand that when using the "Decision Tree Learner" the user can modify the parameters and control in some extend the variables that build the tree. However want to know why the decision tree discarded or dropped some variables. 

For instance I want to predict Y and the independ variables are X1, X2, X3, X4... and the tree that I built just selected the X1 and X4 variables and dropped X2 and X3. 

One idea is to make a correlation matrix, but X2 is a string variable. 

I wonder if I can use the workflow used by Iris in this post "Variable Importance in Prediction (Classification or Regression) Molels" to see importance of that variable. 

Thank you