Refer to Exercise 19 for a description of the data set. a. Create a random forest ensemble

Question:

Refer to Exercise 19 for a description of the data set.

a. Create a random forest ensemble classification tree model. Select three predictor variables randomly to construct each weak learner. What are the overall accuracy rate, sensitivity, and specificity of the model on the validation data? What is the AUC value of the model? Which is the most important predictor variable?

b. Compare the performance of the random forest ensemble model to that of the single-tree model created in Exercise 19 (for R) or Exercise 20 (for Analytic Solver). Which model shows more robust performance? Explain.

c. Score the new cases in the Graduate_Score worksheet using the random forest ensemble classification tree model. What is the probability of the first student graduating within four years? What is the probability for the second and third students?

Data from Exercise 19

Dereck Anderson is an institutional researcher at a major university. The university has set a goal to increase the number of students who graduate within four years by 20% in five years. Dereck is asked by his boss to create a model that would flag any student who has a high likelihood of not being able to graduate within four years. He has compiled a data set of 2,000 previous students of the university that contains the following variables: sex (M/F), whether the student is Caucasian (White), high school GPA (HS GPA), SAT score (SAT), College GPA (GPA), whether the student’s parents are college educated (College Parent), and whether the student graduated within four years (Grad). A portion of the Graduate_Data worksheet is shown in the accompanying table. Build a default classification tree to predict whether the student will be able to graduate within four years (Grad). Display the classification tree.