Data Visualization and Dashboarding
Data Visualization and Dashboarding
This domain tests chart choice, encoding accuracy, and dashboard communication. Expect questions that show a chart type or design decision and ask what it communicates, what it distorts, or what should replace it.
Core rules
- Match chart to task: histogram or box plot for distribution, bar chart for categorical comparison, scatter for relationships, line for trend over time, stacked bar for composition.
- Label axes with units and use readable scales; avoid unnecessary dual axes.
- Use sequential color for ordered values, diverging color for signed values, and categorical color for groups. Prefer colorblind-safe palettes.
- Dashboards should answer one question per view, show hierarchy, and provide filters without clutter.
- matplotlib is the foundation; seaborn simplifies statistical plots; plotly and Dash add interactivity; folium handles maps.
Worked example
A stakeholder asks whether ad spend predicts revenue across 40 stores. Use a scatter plot with spend on the x-axis and revenue on the y-axis, and add a regression line with seaborn regplot. Reasoning: both variables are continuous, so a relationship plot is appropriate. The regression line shows direction and approximate strength, while annotated outliers invite follow-up. A bar chart of average revenue by spend bucket would hide spread and create false precision.
Traps
- Pie charts with many slices or similar values.
- Truncated y-axes that exaggerate small differences.
- Rainbow colormaps for continuous data.
- Unordered categories on a bar chart.
- Missing units or unlabeled axes.
- Implying causation from a correlation plot.