Choosing a chart¶

The library exposes 15 chart types. The right one depends on the question you want the terminal output to answer.

Comparison charts¶

Use these when the reader needs to compare magnitudes or relative change.

Chart	Use when
`BarChart`	Comparing independent values across categories
`ComparisonBar`	Showing planned vs actual or before vs after values
`DivergingBar`	Highlighting positive and negative change around a center
`StackedBar`	Showing composition of a total by segment
`NormalizedSpeedup`	Comparing speedup ratios in performance analysis

Distribution charts¶

Use these when spread, skew, or percentile behavior matters.

Chart	Use when
`Histogram`	Showing bucketed frequency counts
`BoxPlot`	Showing quartiles, whiskers, and outliers
`PercentileLadder`	Comparing percentile cut points across groups
`CDFChart`	Showing the cumulative shape of a distribution

Trend and relationship charts¶

Chart	Use when
`LineChart`	Showing ordered values over time or sequence
`ScatterPlot`	Showing correlation such as ad spend vs conversions
`SparklineTable`	Embedding tiny trends inside a textual table

Tabular and matrix output¶

Chart	Use when
`Heatmap`	Showing dense matrix intensity at a glance
`RankTable`	Showing rank-ordered summaries across criteria
`SummaryBox`	Showing headline metrics and deltas in one panel

Preferred API¶

For new code, use the clean exported names and their data models:

BarChart with BarData
Heatmap with matrix and labels
LineChart with LinePoint
SummaryBox with SummaryStats

Performance-specific note¶

NormalizedSpeedup is the chart type most closely tied to benchmark and performance-analysis workflows. ComparisonBar, DivergingBar, PercentileLadder, RankTable, and SummaryBox are still general purpose charts, but they also map naturally to performance reporting.