Mini-Project 2

Overview

Choose a dataset, import and clean it, explore it with transformations and visualizations, and communicate what you find in a rendered Quarto document. You will also present your work and describe one piece of code in detail.

Data Options

The following datasets are listed on kaggle.com. These are suggestions only. You may use another dataset (including the Zillow data from class) with instructor approval.

Work & Wellness

Environment

Requirements

Data import and cleaning

Import the data with a reproducible script or chunk (e.g., readr::read_csv()); no manual edits to the raw file.
If you download data, do it in code (dir.create(), download.file()) so the project can be re-run.
Clean variable names (e.g., janitor::clean_names()) and check variable types.
Investigate missing values: how many, where, and how you handled them.
Do not leave stray objects around; start from a clean session when rendering (rm(list = ls()) is not needed in a .qmd, which renders in a fresh session).

Transformations

group_by() + summarize(), and actually use/plot the results
select(), mutate(), distinct(), filter()
Pivoting of some kind (pivot_longer() or pivot_wider())
Optional: a join (e.g., left_join()) with a second table such as state or region lookups

Visualizations

At least three plots using ggplot2, each with a purpose.
At least one plot using geom_text() or geom_label().
Appropriate choices: line plots for change over time, bars for counts or aggregates
Informative titles, axis labels, and legends (labs()); readable scales (scales::label_comma(), label_dollar(), etc.).
Optional/encouraged: a map, a divergent bar chart, or animation

Statistics

Use some language of basic statistics.
State whether you are using descriptive statistics or inferential statistics, and why.
Report summaries with context (e.g., mean/median, spread, n). If you run an inferential test, state its assumptions and check them.
Use language proportional to the evidence. Avoid claiming causation from observational data.

Coding and Quarto

Good, brief comments in the .qmd, explaining why rather than narrating what.
Code chunks have labels, and figures have captions or alt text (#| label:, #| fig-cap:, #| fig-alt:).
Use the base pipe |> consistently.
Rendered with Quarto to a self-contained .html file.
Use execute: options to hide warnings and messages, and decide deliberately on echo / code-fold.
Optional: inline code (`r`) to report numbers in text, and cross-references to figures (@fig-...).

Presentation

5 minutes: dataset, question, one key finding.
Describe in detail one piece of code (what it does, line by line, and why you wrote it that way).
Optional: present as a revealjs slide deck.

Rubric

Category	Points	Full credit
Data import and cleaning	15	Reproducible import; names and types checked; missing values investigated and handled with justification
Transformations	20	All of `group_by`/`summarize` (and used), `select`, `mutate`, `distinct`, `filter`, and pivoting present, correct, and purposeful
Visualizations	20	At least 3 clear plots, including `geom_text`/`geom_label`; labeled, readable, appropriate chart types
Statistics	10	Descriptive vs. inferential stated; summaries interpreted correctly; assumptions noted where relevant
Code quality and comments	10	Clear, consistent style; useful comments; labeled chunks; runs top to bottom
Quarto document and publishing	10	Renders without errors to a self-contained `.html`; organized with headings; figures captioned
Narrative and interpretation	5	Clear question and conclusions supported by the results; limitations acknowledged
Presentation	10	Clear and on time; one piece of code described in detail
Total	100

Performance levels (per category)

Level	Share of points	Description
Excellent	90-100%	Complete, correct, and thoughtful; goes beyond the minimum
Proficient	75-89%	Meets requirements with minor errors or gaps
Developing	50-74%	Some requirements missing or incorrect
Incomplete	< 50%	Major requirements missing, or document does not render

Submission

Submit the rendered .html file and the .qmd source.
Due date: tentatively Oct. 21.