Representing Data

I took the “Representing Data” course with two primary motivations: developing data visualization expertise and exploring the intersection of ethics and visualization. I found myself reflecting on how I had previously accepted datasets without question, focusing instead on aesthetic presentation.

Course Structure and Tools

The course teaches data visualization through a deliberate progression of technologies:

1. Sketches

The foundational step requiring deliberate decisions about what to visualize. A key insight: data visualization does not mean you have to visualize every single column and row. Data stories often represent subsets of complete datasets, similar to how films capture only portions of total footage.

2. RawGraphs

A tool for rapid experimentation allowing users to load datasets and test visualizations without constant reconfiguration. Pre-configured visualizations require only column assignment.

3. Tableau

The most robust application in its category, though with a steep learning curve. Our group project utilized Tableau, demonstrating quick iteration capabilities, though nuanced visualizations demand significant commitment.

4. Python

For statistical analysis work, focusing on Seaborn, Altair, and Vega-Altair packages. Python offers greater control than Tableau for specialized use cases like ML model performance visualization.

Application to Thesis

The course prompted reconsideration of my thesis on shifting ethics from compliance to competency. Rather than static tables, I began considering creating interactive visualizations or even structuring the entire project as a Jupyter notebook --- prioritizing practical integration into developers’ workflows rather than purely academic presentation.