What This Document Is
This document contains solutions to Homework Eight for Northwestern University’s Statistical Methods For Data Mining (IEMS 304) course. It focuses on applying regression tree and nearest-neighbors methods to a prostate cancer dataset, and neural networks.
Why This Document Matters
This assignment solution is intended for students enrolled in IEMS 304 who have completed Homework Eight. It serves as a reference for checking understanding of the course material and validating implemented solutions. It’s used after attempting the assignment to review correct approaches and identify areas for improvement.
Common Limitations or Challenges
This document provides completed solutions; it does not offer explanations of the underlying statistical concepts or guidance on *how* to arrive at those solutions. Students still need to understand the principles of regression trees, nearest neighbors, and neural networks to fully benefit from the material.
What This Document Provides
The full document includes:
* R code and output for fitting and evaluating regression trees using the `tree` package.
* Calculations and results for nearest-neighbors predictions, both manually in Excel and using R.
* R code and results for fitting a neural network model using the `nnet` package, including parameter tuning via cross-validation.
* Specific predicted values for given input data points.
* Discussion of variable importance based on the fitted models.
This preview does *not* include the detailed R code, Excel spreadsheets, or step-by-step derivations of the solutions. It only describes the types of problems addressed and the methods used.