Homework Assignment

Please complete the following tasks within one week and submit your work as a GitHub repository.

---

1. Extend the k-NN S3 Class (30 pts)

• Different Distances: Update your knn_s3 implementation to support a dist = c("sse", "sad") argument for L1 and L2 distances.
• Documentation: Add roxygen2 comments for the new argument and regenerate the help files using devtools::document().

---

2. Shiny App Enhancement (30 pts)

• Dynamic Predictors: In the k-NN app, add a multi-select input that allows users to choose predictor variables at runtime.
• Download Button: Implement functionality to download the test-set predictions as a CSV file.

---

3. Rcpp Performance Study (30 pts)

• Benchmark Script: Create benchmark_knn_dims.R to compare runtimes for pure R vs Rcpp implementations across combinations of sample size \(n \in \{500, 1000, 2000\}\) and feature count \(p \in \{3, 5, 10\}\).
• Summary Report: Produce a one-page PDF summarizing your benchmarks, key code changes, and performance recommendations.

---

Deliverables

Your GitHub repository should be organized as follows:

knn_workshop_hw/
├── R/           # R scripts and tests
├── app/         # Shiny app files
├── src/         # C++ source for Rcpp
├── benchmark_knn_dims.R
├── report.pdf   # summary of benchmarks and findings
└── README.md    # overview and instructions

The README must include: - Setup instructions for all components - A sample run demonstrating each part of the assignment

---

Good luck, and feel free to reach out with any questions!