Tableau Chart Types & Visualization Selection¶

Choosing the right chart type is a core BI skill. Each visualization encodes data attributes differently, and the choice should be driven by the specific question being answered. Anscombe's Quartet demonstrates why: four datasets with identical statistics produce completely different visual shapes.

Key Facts¶

Visualization = set of encoding methods mapped to data
For precision (exact values): use length and position on axes
For separation (grouping): use color, volume, size
Rule: understand what question the user is asking, then choose the chart that answers it
Selfmade visualizations = internal analysis (Jupyter, matplotlib); Premade = communication to users (dashboards, presentations)

Patterns¶

Data Encoding Attributes¶

Main encoding attributes ranked by precision: 1. Position on common scale (most precise) 2. Length 3. Angle/slope 4. Area 5. Color intensity 6. Color hue (least precise for quantitative data)

Chart Type Selection Guide¶

Chart	Best For	Notes
Bar/Column	Comparisons between categories	Horizontal bars for long labels
Line	Trends over time	Connect only continuous data
Pie/Donut	Part-to-whole	Use sparingly, max 5 slices
Matrix/Pivot	Cross-tabulation	When precise numbers matter
Card	Single KPI number	Dashboard header factoids
Table	Raw tabular data	Detail drill-down
Treemap	Hierarchical part-to-whole with size	Better than pie for many categories
Map/Filled Map	Geographic data	Check color interpolation
Gauge	Progress to goal	Single-metric target tracking
Scatter	Correlation between two measures	Add trend line for clarity
Heatmap	Pattern density across two dimensions	Sequential color scale

Visualization Categories¶

Selfmade (for internal analysis): - Business analytics: quick data exploration (Jupyter + matplotlib/seaborn) - Scientific visualization: physical processes, 3D - Search/concept visualization: relationships and principles

Premade (to communicate to users): - Dashboards and presentations: interactive business performance panels - Entity cards / personal accounts: banking-style views (KPI factoids + sparklines) - Data analysis tools: user explores data themselves via interactivity - Infographics and journalism: attention-grabbing, design-heavy

Sparklines and KPI Factoids¶

KPI factoid: single number with label, shows current state (e.g., "Sales $5.7M")
Sparkline: tiny inline chart showing trend without axes, placed next to factoid
Both placed in upper-left area for maximum attention and quick state assessment

Dashboard Actions (Interactivity)¶

Action	Description
Filter	Click element -> filter other charts (most common)
Highlight	Click -> highlight related data without filtering
Parameter	User action -> change parameter value
Set	User action -> add/remove from set
GoToSheet	Navigate to another worksheet/dashboard
URL	Open URL with dynamic parameters

URL action example:

https://crm.example.com/deal?object=<object_number>

Gotchas¶

Anscombe's Quartet: always visualize data before trusting summary statistics - four datasets with identical mean, variance, correlation, and regression line look completely different
Pie charts with more than 5 slices become unreadable - use bar chart or treemap instead
Large colorful objects attract attention regardless of position (contrast overrides F-pattern reading)
Incorrect color interpolation in heatmaps and maps makes charts misleading - always verify
Each worksheet generates one query to the source - more charts = more queries = slower dashboard