Regression risk matrix

Fission

Product

▾

Setup

›

›

Regression risk matrix

Regression risk matrix uses Fission Charts typed Rust data to render a production-ready heatmap view. Use it before releases to focus attention. The screenshot is captured from the native Fission chart gallery.

What the chart is for

Regression risk matrix is part of the Heatmap chart family and is intended for Risk values by component and area. Use it when the visual form makes the user's question faster to answer than a table or a simpler chart would.

Avoid it when the visual form makes the user estimate more than necessary; choose the simplest chart that answers the product question.

Data model

Risk values by component and area. Keep the data close to the Rust type that describes it. Fission Charts is typed Rust, so each chart uses explicit series and component structs instead of string configuration hidden in a loosely typed object.

Rust API

Field	Type	Notes
title	&str	Names the chart for the screen, accessibility tree, and test output.
x_axis	Axis	Usually Axis::category; use Axis::value() when x positions are numeric.
y_axis	Axis::value()	Maps each sample value to the chart range.
series	Series	Provides the typed data model for this chart family.
width / height	f32	Optional fixed size; omit them when the chart should flex inside Fission layout.

Example

use fission::charts::{Axis, Chart, HeatmapSeries, VisualMap};

let chart = Chart::new()
    .title("Load")
    .x_axis(Axis::category(vec!["Mon", "Tue", "Wed"]))
    .y_axis(Axis::category(vec!["AM", "PM"]))
    .visual_map(VisualMap::new().min(0.0).max(10.0))
    .series(vec![HeatmapSeries::new("Load")
        .data(vec![(0, 0, 3.0), (1, 0, 8.0), (2, 1, 5.0)])
        .into()]);

Interaction and animation

Charts can emit typed ChartInteractionEvent values when interaction is enabled. Handle those events in a reducer when the app needs hover, press, release, scroll, selection, or brush behavior. ChartAnimation stores duration, delay, stagger, easing, and reduced-motion behavior as deterministic chart data, so animation timing can be tested instead of being hidden in ad-hoc timers.

Testing guidance

For this chart, test the data mapping first, then test lowering, then capture a live screenshot when visual output changes. The screenshot for this page is refreshed by running the ignored chart-gallery screenshot capture test with FISSION_UPDATE_CHART_DOCS=1, which updates documentation/static/img/charts before the static site build.

Authoring from a Fission component

Use this reference page as the chart expression inside ordinary Fission component conversion. The chart is a Widget, so it can sit inside a Card, Grid, Scroll, responsive page section, or any other layout container.

use fission::prelude::*;
use fission::charts::{Axis, Chart, HeatmapSeries, VisualMap};

pub struct RegressionRiskMatrixChart;

impl From<RegressionRiskMatrixChart> for Widget {
    fn from(_: RegressionRiskMatrixChart) -> Widget {
        Chart::new()
            .title("Load")
            .x_axis(Axis::category(vec!["Mon", "Tue", "Wed"]))
            .y_axis(Axis::category(vec!["AM", "PM"]))
            .visual_map(VisualMap::new().min(0.0).max(10.0))
            .series(vec![HeatmapSeries::new("Load")
                .data(vec![(0, 0, 3.0), (1, 0, 8.0), (2, 1, 5.0)])
                .into()])
            .into()
    }
}

Keep expensive data loading outside component conversion. A reducer, job, service, or server route should prepare the typed chart data, then the component should read that state and construct the chart deterministically.

Options, accessibility, and diagnostics

Area	What to decide	How to verify
Data shape	Keep source rows in typed Rust structs, then map them into the series type shown in the example.	Unit test the mapping separately from rendering.
Options	Choose axes, legends, labels, animation, and interaction based on the user's task.	Add a screenshot test when changing visual behavior.
Accessibility	Provide a clear title and adjacent summary text for important trends or outliers.	Inspect the generated semantics and make sure the chart is understandable without color alone.
Failure handling	Render an empty, loading, or error state before constructing the chart if data is unavailable.	Test empty data, partial data, and failed fetches.
Performance	Prefer summarized or windowed data for very large datasets; keep full raw history in the data layer.	Profile frame time and interaction latency with representative data volumes.