Assignment 5: SAS - Practice with Ensemble Models

Make a presentation of you doing the follows. Using Zoom to record a video of the presentation and the link of the video or the video to Canvas.

Video Instruction

  1. Create a SAS Library and into the library the following dataset

Breast Cancer Wisconsin

  1. Create a data source from the above dataset for model building. Reject all the text variables. Set diagnosis as the binary target.

  2. Split the data 70:30:0 for Training: Validation: Test. Train the following models.

    • A random forest of 100 trees maximum

    • A random forest of 200 trees maximum

    • A random forest that have 150 trees maximum and consider 3 variables at each split for the trees.

    • A gradient boosting with 200 iterations and the learning rate (shrinkage) of 0.05

    • A bagging model of 5 regression models

    • A boosting model of 5 regression models

  3. Out of the models in Question 3, what is the best model in term of misclassification?

  4. Make an effort to find a better model than the model in Question 4