Visualization requirements
Complete the fields that best describe your dataset.
Recommendation results
Chart preview
Export report
Generate a recommendation first, then export the results.
Available chart library
How to choose a data visualization type
Choosing a visualization starts with the question you want answered. Comparison questions usually need bars because lengths are easy to compare. Trends over ordered time periods normally work best with line charts. Distribution questions require plots that reveal spread, center, and unusual values. Histograms and box plots often handle that task well.
Match charts to data structure
The structure of your dataset matters as much as the goal. Two numerical variables often point toward a scatter plot. Categories with one measure usually fit a bar chart. Percentages forming one total can use a donut or pie chart when categories remain few. Many categories are clearer in horizontal bars. Hierarchical data can use treemaps or sunburst charts. Geographic values may require choropleth maps or symbol maps. Dataset size also changes the practical choice. Large datasets benefit from aggregation, filtering, or interactive exploration. Smaller datasets can support direct labels and richer annotations. When several measures share categories, grouped or stacked displays may communicate patterns more efficiently.
Consider audience and precision
Audience expectations influence complexity and styling. Executives often need quick comparisons and obvious takeaways. Technical users may accept dense plots, confidence intervals, or matrix views. Presentation charts should remain readable from a distance. Mobile dashboards need compact layouts and limited labels. Accessibility also matters. Color should not be the only way to distinguish groups, and important labels should remain readable.
Avoid misleading choices
A visually attractive chart can still communicate poorly. Pie charts become difficult with many slices. Three-dimensional effects distort perceived values. Dual axes can imply relationships that are not meaningful. Stacked charts make middle segments hard to compare. A chart selector should therefore explain both recommended and unsuitable choices. This tool combines analytical intent, data type, variable count, density, medium, and accuracy needs. The final recommendation provides a best choice, alternatives, suitability score, usage notes, and warnings. These checks make visualization decisions faster and more consistent.
Common visualization decisions
| Goal | Typical data | Strong choices | Common caution |
|---|---|---|---|
| Compare categories | Categorical + numerical | Bar, dot, lollipop | Avoid unnecessary 3D bars |
| Show a trend | Date/time + numerical | Line, area | Keep time order consistent |
| Show distribution | Numerical | Histogram, box, violin | Choose sensible bins |
| Show relationship | Two numerical variables | Scatter, bubble | Do not imply causation |
| Show composition | Categories summing to a whole | Stacked bar, pie, donut | Limit pie slices |
| Show geography | Location + measure | Choropleth, symbol map | Normalize when appropriate |