← All Data Visualization with AI modules

Capstone — AI-Assisted Statistical Data Story

Graduate defense · R/Quarto or Shiny artifact · verified AI workflow

Build a reproducible statistical visualization project that answers a meaningful question, verifies every AI-assisted claim, and defends the visual/statistical choices.

Required evidence

Capstone readiness tracker

Data question is statistically answerable and includes an inference boundary. Dataset is documented with a data dictionary and cleaning log. Missingness, sampling, and measurement limitations are addressed. At least one distribution/comparison visual is verified. At least one uncertainty/model/relationship visual is verified. R/Quarto report, R script bundle, or Shiny app source is included. AI prompt log documents accepted/rejected/corrected output. Final interpretation states evidence, assumptions, uncertainty, and limits.

Submission formats

Option A

Quarto report

A reproducible HTML/PDF report with narrative, R code, outputs, figures, AI log, and verification appendix.

Option B

Shiny app

An interactive dashboard or statistical explorer with source code, metric definitions, and written statistical notes.

Option C

GeoGebra-enhanced visual explanation

An interactive math visualization paired with AI prompt logs, R verification, and a written defense of what the construction proves and does not prove.

Option C

R bundle

Documented R scripts, exported visuals, and a defense memo if Quarto/Shiny is not appropriate.

Rubric

Category	Points
Data question, dataset fit, and statistical framing	15
Cleaning, missingness, and reproducibility	15
Statistical analysis: summaries, uncertainty, model/relationship reasoning	20
Chart choice, visual encoding, and accessibility	15
AI use documented, audited, and verified	15
Explanation/story clarity and mathematical defense	10
Ethics, limitations, and inferential boundaries	10

Defense prompts need JavaScript.

Passing standard A score of 70/100 earns the credential, but graduate-ready work should be able to survive a statistical defense: why this data, why this visual, why this model or summary, what uncertainty remains, and exactly how AI was verified.