Charts
The datachart.charts module of the datachart package provides various chart types to create data visualizations. The module is designed to be highly customizable and easy to use.
The charts are grouped by the question they answer. Each card names a chart, says what it is for, and links to its how-to guide. The chips under the name say how the chart composes with the other charts through the composition functions:
- Panel — the chart can be overlaid with other charts in one coordinate space through Panel. A chart that owns its whole axes (the heatmap, Sankey chart, treemap, network chart, calendar heatmap, and scatter matrix) or draws a mirrored or task axis (the pyramid and gantt charts) cannot, and shows Panel. The box, violin, raincloud, and ridgeline plots overlay with other kinds of charts, but a panel holds one dataset of each of these kinds.
- Grid — the chart can take a cell of a combined figure through Grid. Every chart can.
A finished chart goes further through the composition guides, the styling guides, and the utility guides.
Trends and Comparisons
Values along an axis or across categories: how a quantity moves and how the categories compare.
-
How a value moves along a continuous axis, and how the trajectories of several series compare.
-
How a total splits into parts along an axis: the top edge traces the total, the bands its composition.
-
Rank over time: who leads and who overtakes whom, with each series named at the end of its line.
-
A numeric value across a few categories; several series can be grouped, stacked, or overlaid.
-
Panel Grid
Two groups mirrored over the same ordered categories, such as an age-sex population pyramid.
-
Series on polar axes: a radar profile over several metrics, or bars over cyclic categories.
-
Panel Grid
One cell per day, weeks as columns, for a daily series with a weekly or seasonal rhythm.
-
Panel Grid
A schedule: one bar per task over a date axis, with groups, progress, milestones, and dependencies.
-
Two values per category and the gap between them: before and after, or a minimum and a maximum.
Distributions
The spread of the values within each group, from a binned summary to every observation.
-
The shape of one numeric variable, binned into counts; a few distributions can be overlaid.
-
The median, quartiles, whiskers, and outliers per group, for comparing many groups compactly.
-
The density profile of each group, showing the skew and the modes that a box plot hides.
-
Every observation as a point at its group, spread so that none hide; for small to medium samples.
-
The density, the raw observations, and the quartile box of each group, side by side in one view.
-
One density ridge per group, stacked and overlapping: how a distribution shifts across many groups.
Relationships
How two or more variables relate to each other.
-
Two numeric variables per observation, with an optional regression line and correlation coefficient.
-
Panel Grid
Every cell of a matrix as a color: correlations, confusion matrices, feature-by-time tables.
-
A surface sampled on a grid, as iso-lines or filled bands: densities, loss landscapes, terrain.
-
Point density on hexagonal tiles, where a scatter chart would turn into an opaque blob.
-
Each record as a polyline across one axis per dimension, colored by group to compare the groups.
-
Panel Grid
Relational data as a node-link diagram: edge weight sets the width, node group sets the color.
-
Panel Grid
A scatter chart for every pair of dimensions, with each dimension's distribution on the diagonal.
Flows
How a quantity moves between categories: where it comes from and where it goes.
-
Panel Grid
Weighted flows between categories through ordered stages: funnels, label transitions, energy budgets.
Part of a Whole
How a whole splits into parts, and parts into smaller parts.
Across Charts
The options every chart shares, worth a guide of their own.
-
Emphasizing and muting data series with the
emphasisoption, by role or by rule, on any chart.