Good charts answer a question—label axes, show uncertainty when relevant, avoid chartjunk, and match chart type to data type (time series → line, distribution → histogram/boxplot).
Checklist
- Clear title and axis labels with units
- Readable font sizes for slides vs papers
- Honest y-axis (avoid truncating unless intentional)
- Document data source and filters (SQL export date, etc.)
Common mistakes
- Pie charts with many slices
- Dual y-axes without careful interpretation
- Overplotting thousands of points without alpha or aggregation
Aggregate in SQL when datasets are huge, then visualize samples in R.
Important interview questions and answers
- Q: When use bar vs line?
A: Bars for categorical counts; lines for ordered/time continuous x. - Q: Why document filters?
A: Reproducibility—readers must know population definition matching Data Science governance.
Self-check
- Name one axis labeling requirement.
- When aggregate before plotting?
Tip: Aggregate huge tables in SQL before ggplot—millions of points overplot without alpha or bins.
Interview prep
- Label axes?
Include units and clear descriptions—audiences should read charts without guessing.