From Data to Thesis: Research Data Analysis with R
Playground, Chapter 18: Using AI with R

How to use this project
1. Double-click chapter18.Rproj. RStudio opens with this folder as the working directory.
2. Open exercises.R. It has four parts:
   A. Practise the chapter       the book's exercises, with blanks (______) to fill
   B. Go further                 new exercises beyond the book
   C. Check your understanding   short questions to answer in your own words
   D. Do it yourself             open tasks: no starter code and no answers
   Run a line with Ctrl+Enter (Cmd+Enter on a Mac).
3. Check parts A to C in solutions.R, after you have tried each one yourself.

The exercises need these packages (install once):
install.packages(c("dplyr", "stringr", "ellmer", "usethis"))

Everything except Exercise 4 and Task 2 runs without a language model.
The chapter uses a free local model, which needs no key and keeps the data on
your computer:
- Install Ollama from ollama.com, then run this once in a terminal:
    ollama pull gemma4:e4b
  (a download of about 10 GB; a computer with a graphics card is much faster).
Or use an online model: set use_local <- FALSE in the scripts, create an API key
with a provider (for example Anthropic), run usethis::edit_r_environ(), add a
line such as ANTHROPIC_API_KEY=your-key, save, and restart R. Never put a key
in a script. Real research data may only be sent to an online service if your
ethics approval allows it.

Files
- exercises.R               the four parts, with starter code for parts A and B
- solutions.R               solutions for parts A and B, model answers for part C
- students.csv              one row per student in the wellbeing study (600 students)
- semesters.csv             one row per student per semester
- open_responses.csv        the students' answers to the open question
- open_responses_coded.csv  200 answers hand-coded into seven themes
- ai_coding_results.csv     the chapter's saved language-model coding (gemma4:e4b)

Please note: Elaf and her data are fictional. The data was generated by a
computer program to look realistic, and its patterns were built in by the
author. It describes no real people and is not evidence about real students;
do not cite or use it as research findings.
Browser exercises for this chapter are at:
https://polla-fattah.github.io/R4NTR/playground/chapter18.html
