How to do it?:
Open the Rmarkdown file of this assignment (link) in Rstudio.
Right under each question, insert a code chunk (you can use the hotkey Ctrl + Alt + I to add a code chunk) and code the solution for the question.
Knit the rmarkdown file (hotkey: Ctrl + Alt + K) to export an html.
Publish the html file to your Githiub Page.
Submission: Submit the link on Github of the assignment to Canvas
Use the Adult Census Income dataset. We will predict the income (whether or not it is more than 50k or not) of an adult. Import the dataset. Partition the data into 80% training and 20% testing.
Practice Decision Tree. Do the follows:
Use rpart package, create a decision tree with maximum depth of 3.
Calculate the accuracy of the model on the testing data. Notice that the positive outcome here is not 1 but >50K or <50K.
Plot the tree
Plot the variable importance by the tree
Create 3 more trees and compare the testing accuracy of these trees, which tree give the highest testing accuracy.
Practice Random Forest. Do the follows:
Use randomForest package, create a random forest of 1000 trees.
Calculate the accuracy of the model on the testing data.
Plot the variable importance by the forest
Create 3 more forests and compare the testing accuracy of these forests, which forest give the highest testing accuracy.
What is the best model (in term of testing accuracy) among all models (including trees and forests) you have trained?