Topic 7 | Finishing Your EDA and Data Descriptor

Class Details

Date: May, 2026
Synopsis: Catching up on pending work, completing the Quarto EDA report, and advancing the data descriptor manuscript.

Class overview: This class is dedicated to finishing what was started in classes 4–6. By the end of this session you will have: a complete, rendering Quarto EDA report and a near-complete data descriptor manuscript.

---

Timing guide

Two groups (Master’s students and PhD students) run independently, same structure.

Segment	Duration
Progress check and individual priorities	20 min
Guided work session	80 min
Final commit and wrap-up	20 min
Total	~120 min

Progress Check and Individual Priorities

Open your project, (pull from GitHub only if you are using the repository), and take stock of where you are. Render your eda_report.qmd from a clean R session to confirm it works.

Together with the instructor, identify what is still pending. Common items at this stage:

• Tidying script incomplete or producing warnings.
• Quarto report missing explanatory prose between code chunks.
• Plots not yet polished (labels, units, color palettes).
• Data descriptor sections still empty or too vague.

Write down your priorities for this session. You have 80 minutes of guided work time: Use them on whatever brings your project closest to completion.

Checkpoint

Every student should be able to state, in one sentence, what they need to finish today. If you are unsure, that is exactly what this check is for.

---

Guided Work Session

This is hands-on time. Work on your own project; the instructor circulates to help with individual issues.

Use the checklist below to track your progress. By the end of this session, every item should be done or close to done.

Quarto EDA report

• The report renders from a clean R session without errors.
• Every code chunk has a short paragraph explaining the purpose and the finding.
• All plots have descriptive labels, units, and a color-friendly palette.
• Summary statistics are computed per group where appropriate.
• code-fold: true is set in the YAML header.

Data descriptor manuscript

• Title describes the dataset, not the research question.
• Abstract summarises what the dataset contains and how it was collected.
• Background and Summary provides scientific context.
• Methods describes data collection and processing (reference your tidying script).
• Data Records lists all variables with names, types, units, and meaning of missing values.
• Technical Validation references your EDA findings (distributions, outlier checks, group balance).
• Code Availability points to your GitHub repository.

Note

If you finish early, review your README file: Does it still accurately describe the current state of the project? Update it if needed.

---

Final Commit and Wrap-up

Commit your work (if you are using Git and GitHub)

1. Stage all updated files.
2. Write a descriptive commit message.
3. Commit and push to GitHub.

What was accomplished across classes 4-7

Over four classes, you built a complete reproducible analysis project:

• Class 4: You presented your data, learned from professional data descriptions, and structured your dataset into tidy format.
• Class 5: You verified the tidy data and explored it through summary statistics and visualisation.
• Class 6: You polished your plots, created a Quarto report, and began the data descriptor.
• Class 7: You completed the Quarto report and data descriptor manuscript.

Your GitHub repository now should contain: Raw data, a tidying script, an EDA and plotting script, a Quarto report that renders to HTML, polished figures, and a data descriptor manuscript. Every step is documented, version-controlled, and reproducible.

Homework

1. Render the final version of eda_report.qmd from a clean R session and confirm it runs without errors.
2. Review the data descriptor for completeness: every section should have content, every variable should be described, and the GitHub repository should be referenced.
3. Final commit and push. Your repository should be in a state where someone unfamiliar with the project can clone it, read the README, render the report, and understand the entire analysis.

---