Safety Shift Plot: test evidence

Requirement-traced qualification evidence for the safety.viz shift-plot module.

Requirement matrix ↗ — the reviewed source specification these tests trace to.

Scope: 28 coverage rows38 distinct requirement IDs
Tests executed: 217 automated checks36 browser · 181 unit
Result: pass all passing9 evidence screenshots
Generated: 2026-08-15 09:05 UTC
Environment: linux 6.17.0-1022-azure · node v22.23.2 · playwright 1.61.1 · chromium 149.0.7827.55
Test run: Actions run #31876074666

Scope & approach

Traceability for the shift-plot module (#14) — the second full renderer, built on the framework proven by the histogram (#2) — per the convention in CONTRIBUTING.md. Requirement IDs are the SSP-* rows of the 39-row reviewed matrix at safety.agent docs/requirements/safety-shift-plot.md; a few module-scheme IDs (SSP-CHART-002, the module-API SSP-API-001 note) name behavior the matrix does not enumerate. Development follows red-green TDD: matrix row → failing test → minimal implementation.

Each table row traces one requirement to the automated test(s) that evidence it: the Requirement column shows the reviewed requirement text and its ID, the source-matrix rows link back to the specification, the issue column links the implementing work, and the result column shows the recorded outcome of every matching test from the committed evidence.json with its captured screenshots. Browser evidence is captured at fixed conditions (1280×800, device scale 1) on the canonical Linux CI environment.

Browser evidence (Playwright — `tests/e2e/shift-plot.spec.js`)

17 requirement rows · 16 tests

Requirement	Source matrix rows	Issue	Tests & evidence
SSP-CTRL-001/SSP-REQ-002/SSP-CTRL-002/SSP-CTRL-003 SSP-CTRL-001 This drop-down menu is used to filter the lab variable that is displayed in the chart. The default measure is set to the first variable in the data set, but the default value can be adjusted to any measure upon the project team's request. Users can select any pre-defined measure in this filter. Selecting a measure will display the data points associated with that measure and will display the visits that the measure is collected at underneath the chart. SSP-REQ-002 This drop-down menu is used to filter the visit data that will be displayed in the chart. The default measure is set to 0, but the default value can be adjusted to any visit(s) upon the project team's request. Users can select one of more pre-defined visit(s) in this filter to be used as the baseline visit(s) in the chart. These values will be displayed on the x-axis. SSP-CTRL-002 This drop-down menu is used to filter the visit data that will be displayed in the chart. The default measure is set to all visits other than visit 0, but the default value can be adjusted to any visit(s) upon the project team's request. Users can select one of more pre-defined visit(s) in this filter to be used as comparison visit(s) in the chart. These values will be displayed on the y-axis. SSP-CTRL-003 The chart can be configured to have any number of data filters. Each specified filter has a drop-down menu that is used to filter the visit data that is displayed in the chart.	SSP-CTRL-001, SSP-REQ-002, SSP-CTRL-002	#14	pass SSP-CTRL-001/SSP-REQ-002/SSP-CTRL-002/SSP-CTRL-003: renders measure, baseline/comparison visit, and filter controls (#14)
SSP-CHART-002	—	#14	pass SSP-CHART-002: the identity line spans a domain shared by both axes (#14)
SSP-COUNT-001/SSP-REG-005 SSP-COUNT-001 Displayed under the header, this statement provides the total number of participants in the data set in integers as well as the percentage of participants shown to one decimal place. In the default view, all participant data is included in the chart, or 100%, unless the project team requests otherwise. When filters are used, the data shown in the histogram is manipulated and the number of participant records shown at the top of the page may increase or decrease, depending on the variable selected. SSP-REG-005 Confirm that the # of enrolled/randomized participants is visible at the top of the page	SSP-COUNT-001, SSP-REG-005	#14	pass SSP-COUNT-001/SSP-REG-005: the participant note reports shown-of-total participants (#14)
SSP-REG-020 SSP-REG-020 Confirm that invalid results are removed from the data rather than measures with any invalid results. Invalid results include missing and non-numeric values. Then confirm that the number of removed results is noted in the console (You can confirm by downloading the data from CAT, updating the STRESN column so a few values are bad. When you reupload the edited data you will see that the number changes appropriately in the console).	SSP-REG-020	#14	pass SSP-REG-020: missing and non-numeric results are dropped with a reported count and note (#14)
SSP-REG-001 SSP-REG-001 Confirm that measure filter is working as expected	SSP-REG-001	#14	pass SSP-REG-001: changing the measure re-pairs the scatter (#14)
SSP-REG-002/SSP-REG-003 SSP-REG-002 Confirm that baseline visit filter are working as expected SSP-REG-003 Confirm that comparison visit filter are working as expected	SSP-REG-002, SSP-REG-003	#14	pass SSP-REG-002/SSP-REG-003: changing baseline and comparison visits swaps the axes (#14)
SSP-CTRL-003 SSP-CTRL-003 The chart can be configured to have any number of data filters. Each specified filter has a drop-down menu that is used to filter the visit data that is displayed in the chart.	SSP-CTRL-003	#14	pass SSP-CTRL-003: applying a filter updates the participant note (#14)
SSP-REG-006 SSP-REG-006 Confirm that hovering over a data point displays subject ID, baseline value, comparison value, change value, and percent of change	SSP-REG-006	#14	pass SSP-REG-006: the point tooltip reports id, baseline, comparison, change, and percent change (#14)
SSP-REQ-003/SSP-REQ-006/SSP-REQ-007/SSP-REG-004/SSP-REG-012 SSP-REQ-003 Clicking and dragging the mouse across the data points will open a table to the right of the chart that contains a detailed listing of participant records. This linked table contains the participant's subject ID, their baseline value for the selected measure, the comparison value for the selected measure, the difference between the baseline and comparison value using an integer and a percentage. When the table appears, the data points selected will appear in a gray box inside of the chart, lightening all of the data points not selected in the box. When a user clicks outside of the gray box, the gray box disappears along with the detailed table. SSP-REQ-006 Create a gray box around the selected data points SSP-REQ-007 Lighten all of the data points not contained in the table SSP-REG-004 Confirm that detailed record can be viewed by clicking and dragging the mouse across a section of values SSP-REG-012 Confirm that the detail listing isn't initially available but then does display when the chart is brushed	SSP-REQ-003, SSP-REQ-006, SSP-REQ-007, SSP-REG-004, SSP-REG-012	#14	pass SSP-REQ-003/SSP-REQ-006/SSP-REQ-007/SSP-REG-004/SSP-REG-012: brushing opens the listing, boxes the selection, and de-emphasizes the rest (#14)
SSP-REG-011 SSP-REG-011 Confirm that unselecting data points resets the detailed record at the bottom of the page	SSP-REG-011	#14	pass SSP-REG-011: clearing the selection resets the points and hides the listing (#14)
SSP-REG-008/SSP-REG-009/SSP-REG-010 SSP-REG-008 Confirm that Search works to filter the data displayed in the listing SSP-REG-009 Confirm that Sort works in the listing by clicking on column headers SSP-REG-010 Confirm that you can export the displayed data to a csv via via the Export: CSV button	SSP-REG-008, SSP-REG-009, SSP-REG-010	#14	pass SSP-REG-008/SSP-REG-009/SSP-REG-010: the listing searches, sorts, and exports to CSV (#14)
SSP-API-003/PPRF-SSP-004 SSP-API-003 The custom `participantsSelected` event is dispatched on the shell root (bubbling, so wrapper/element-level listeners keep working) whenever the details for one or more participants are viewed (or cleared) by brushing the chart. The event has a custom `data` property holding an array of the selected IDs (`["123-456-7"]`) or an empty array when participant details are cleared.	SSP-API-003 + PPRF-SSP-004 (participant-profile matrix)	#14	pass SSP-API-003/PPRF-SSP-004: brushing dispatches participantsSelected on the shell root, bubbling to the element (#14, #99)
SSP-API-001 (module scheme) SSP-API-001 A factory to create a custom webChart chart object.	— (see legacy-API note)	#14	pass SSP-API-001: lifecycle API supports init, setData, setSettings, render, resize, and destroy (#14)
SSP-REG-016/SSP-REG-018 SSP-REG-016 Confirm that the chart appears between controls and listing. SSP-REG-018 Confirm that the controls appear to left of chart.	SSP-REG-016, SSP-REG-018	#14	no record shared shell: controls left of the chart, chart above the listing (`tests/e2e/site.spec.js`)
PPRF-SSP-001	PPRF-SSP-001 (participant-profile matrix)	#99	pass PPRF-SSP-001: a multi-participant brush collapses the rail to a worst-first stepper whose steps emphasize the chart (#99)
PPRF-SSP-002	PPRF-SSP-002 (participant-profile matrix)	#99	pass PPRF-SSP-002: a single-point brush shows the full railed profile with no stepper, beside the linked listing (#99)
PPRF-SSP-003	PPRF-SSP-003 (participant-profile matrix)	#99	pass PPRF-SSP-003: clearing the selection and control-driven redraws empty the rail (#99)

Unit evidence (Vitest — `tests/unit/shift-plot/`)

11 requirement rows · 32 tests

Requirement	Source matrix rows	Issue	Tests & evidence
SSP-CFG-004/005/006 SSP-CFG-004 an object that defines the baseline value SSP-CFG-005 an object that defines the comparison (or post-baseline) value SSP-CFG-006 an array of filter variables and associated metadata	SSP-CFG-004, SSP-CFG-005, SSP-CFG-006	#14	`configure.test.js` pass shift-plot configure SSP-CFG-004/005: baseline/comparison visits normalize to arrays and stats fall back to mean (#14) pass shift-plot configure SSP-CFG-004/005/006: default settings map the standard measure, result, visit, and id columns (#14) pass shift-plot configure SSP-CFG-006: filter specs normalize strings and objects to value_col/label pairs (#14) pass shift-plot structureData SSP-CFG-005: multiple comparison visits collapse with the comparison statistic (#14)
SSP-REQ-005 SSP-REQ-005 Display the raw data records, including the subject identifier, baseline value, comparison value, change, and percent change	SSP-REQ-005	#14	`configure.test.js` pass shift-plot configure SSP-REQ-005: details default to id, baseline, comparison, change, and percent change (#14) pass shift-plot structureData SSP-REQ-005/SSP-REG-019: pairs baseline against comparison per participant with change and percent change (#14)
SSP-REG-020 SSP-REG-020 Confirm that invalid results are removed from the data rather than measures with any invalid results. Invalid results include missing and non-numeric values. Then confirm that the number of removed results is noted in the console (You can confirm by downloading the data from CAT, updating the STRESN column so a few values are bad. When you reupload the edited data you will see that the number changes appropriately in the console).	SSP-REG-020	#14	`structureData.test.js` pass shift-plot structureData SSP-REG-020: missing and non-numeric results are removed with a reported count (#14)
SSP-DATA-001 SSP-DATA-001 Safety Shift Plot is initialized with JSON data files that match the format created by d3.csv(). The chart expects an ADaM-esque data structure by default, but can be customized to work with any dataset that contains one row per participant per time point per measure and includes the required columns specified below.	SSP-DATA-001	#14	`structureData.test.js` pass shift-plot checkInputs SSP-DATA-001: missing required variables throw a clear message (#14) pass shift-plot checkInputs SSP-DATA-001: the JSON schema publishes the data contract with required column mappings (#14) pass shift-plot structureData SSP-DATA-001: measure labels are the raw measure value (#14)
SSP-REG-013/SSP-REG-014 SSP-REG-013 Confirm that the User can specify visit_order_col that names a variable that orders the visits. SSP-REG-014 Confirm that Visit ordering can be set two ways: {"visit_col":"VISIT","visit_order_col":"VISITNUM"} to order visits sequentially and {"visit_col":"VISIT","visit_order_col":"VISIT"} to order visits alphanumerically.	SSP-REG-013, SSP-REG-014	#14	`structureData.test.js` pass shift-plot structureData SSP-REG-013/SSP-REG-014: visits order by visit_order_col, else alphanumerically (#14)
SSP-REQ-005/SSP-REG-019 SSP-REQ-005 Display the raw data records, including the subject identifier, baseline value, comparison value, change, and percent change SSP-REG-019 Confirm that when a measure contains fewer than five unique values it isn't removed from the data (edit the dataset ADBDS so that it contains at least two results for the same measure from the same USUBJID. The shift plot is plotting the baseline value on the x-axis and the comparison value on the y-axis which requires at least two results).	SSP-REQ-005, SSP-REG-019	#14	`structureData.test.js` pass shift-plot configure SSP-REQ-005: details default to id, baseline, comparison, change, and percent change (#14) pass shift-plot structureData SSP-REQ-005/SSP-REG-019: pairs baseline against comparison per participant with change and percent change (#14)
SSP-CFG-005 SSP-CFG-005 an object that defines the comparison (or post-baseline) value	SSP-CFG-005	#14	`structureData.test.js` pass shift-plot structureData SSP-CFG-005: multiple comparison visits collapse with the comparison statistic (#14)
SSP-CTRL-001/SSP-CTRL-003 SSP-CTRL-001 This drop-down menu is used to filter the lab variable that is displayed in the chart. The default measure is set to the first variable in the data set, but the default value can be adjusted to any measure upon the project team's request. Users can select any pre-defined measure in this filter. Selecting a measure will display the data points associated with that measure and will display the visits that the measure is collected at underneath the chart. SSP-CTRL-003 The chart can be configured to have any number of data filters. Each specified filter has a drop-down menu that is used to filter the visit data that is displayed in the chart.	SSP-CTRL-001, SSP-CTRL-003	#14	`structureData.test.js` pass shift-plot structureData SSP-CTRL-001: only the selected measure is paired (#14) pass shift-plot structureData SSP-CTRL-003: active filters subset rows by stringified equality (#14)
SSP-CHART-002	—	#14	`structureData.test.js` pass shift-plot structureData SSP-CHART-002: the shared domain spans both axes with padding (#14)
SSP-DATA-001/SSP-DATA-003 SSP-DATA-001 Safety Shift Plot is initialized with JSON data files that match the format created by d3.csv(). The chart expects an ADaM-esque data structure by default, but can be customized to work with any dataset that contains one row per participant per time point per measure and includes the required columns specified below. SSP-DATA-003 No customization is needed to initialize a chart when the `data` object contains all default variables above. Just pass an empty object (`{}`) for `settings` and initialize the chart like so: `safetyShiftPlot('body', {}).init(data)`.	SSP-DATA-001, SSP-DATA-003	#14	`checkInputs.test.js` pass shift-plot checkInputs SSP-DATA-001: missing required variables throw a clear message (#14) pass shift-plot checkInputs SSP-DATA-001: the JSON schema publishes the data contract with required column mappings (#14) pass shift-plot checkInputs SSP-DATA-003: valid long-format data with optional columns passes (#14) pass shift-plot structureData SSP-DATA-001: measure labels are the raw measure value (#14)
PPRF-SSP-001..004	PPRF-SSP-001..004 (participant-profile matrix)	#99	`profile-adoption.test.js` pass shift-plot cohort stepper (PPRF-SSP-001, PPRF-11) PPRF-SSP-001: a multi-participant brush collapses the rail to a worst-first stepper pass shift-plot cohort stepper (PPRF-SSP-001, PPRF-11) PPRF-SSP-001: stepping emphasizes the stepped point on the chart without re-dispatching pass shift-plot dispatch target (PPRF-SSP-004, SSP-API-003) PPRF-SSP-004: is idempotent under repeated identical dispatches pass shift-plot dispatch target (PPRF-SSP-004, SSP-API-003) PPRF-SSP-004: the dispatch still bubbles to element-level listeners (backward compatibility) pass shift-plot dispatch target (PPRF-SSP-004, SSP-API-003) PPRF-SSP-004: the dispatch travels via the shell root (the house participantsSelected contract) pass shift-plot participant-profile adoption (PPRF-SSP-002) PPRF-SSP-002: a single-point brush shows the full profile with no stepper AND keeps the listing pass shift-plot participant-profile adoption (PPRF-SSP-002) PPRF-SSP-002: derives the rail feed rows ONCE per setData from rawData through the hep-core cleaners pass shift-plot participant-profile adoption (PPRF-SSP-002) PPRF-SSP-002: host listing columns are NOT the rail header details (profile_details defaults empty) pass shift-plot participant-profile adoption (PPRF-SSP-002) PPRF-SSP-002: mounts the rail by default (config-on) into the shell profile slot pass shift-plot participant-profile adoption (PPRF-SSP-002) PPRF-SSP-002: passes the host lab mappings through, mapping visitn_col from visit_order_col pass shift-plot participant-profile adoption (PPRF-SSP-002) PPRF-SSP-002: profile: false leaves the slot empty and mounts no dock pass shift-plot rail clear paths (PPRF-SSP-003, PPRF-11) PPRF-SSP-002: setSettings({profile: false}) unmounts a live dock; re-enabling remounts it pass shift-plot rail clear paths (PPRF-SSP-003, PPRF-11) PPRF-SSP-003: a control-driven render resets the selection AND the rail (render preamble) pass shift-plot rail clear paths (PPRF-SSP-003, PPRF-11) PPRF-SSP-003: clearing the selection (the tiny-click path) empties the rail and the emphasis pass shift-plot rail clear paths (PPRF-SSP-003, PPRF-11) PPRF-SSP-003: destroy tears the rail down with the instance pass shift-plot rail clear paths (PPRF-SSP-003, PPRF-11) PPRF-SSP-003: dock Clear empties an externally-fed cohort the host never selected (PPRF-11) pass shift-plot rail clear paths (PPRF-SSP-003, PPRF-11) PPRF-SSP-003: the rail Clear affordance routes through the host clear path

Visual evidence

Every screenshot below is a committed baseline: the same PNG is the visual-regression baseline the browser suite asserts against and the evidence artifact shown here. Click any capture for the full-resolution image.

PPRF SSP 001 docked stepper
PPRF SSP 001 railed stepper
PPRF SSP 002 docked full profile
PPRF SSP 002 railed full profile
SSP COUNT 001 participant count
SSP CTRL 001 baseline scatter
SSP REG 008 linked listing
SSP REG 020 invalid data note
SSP REQ 003 brushed selection

Source-matrix routing status (39 rows)

Of the 39 reviewed SSP-* rows, 32 are covered by the browser/unit evidence above and 7 are descoped with rationale below (honest routing per the histogram precedent).

Descoped — legacy Webcharts/CAT (5 rows): - SSP-API-001 ("a factory to create a custom Webcharts chart object") describes the legacy Webcharts API, which the designs intentionally do not preserve. The module ships the pilot's lifecycle API instead — SSP-API-001 in the module scheme (see the browser table). - SSP-CFG-001 and SSP-CFG-002 are legacy Webcharts settings passthrough/detail; the nextgen module exposes an explicit settings object (ShiftPlotSettings, covered by SSP-CFG-004/005/006) rather than accepting Webcharts configuration objects. - SSP-DATA-004 is a CAT/viz-library "customized chart with queries" example page; the nextgen demo uses the standalone real-data page instead. - SSP-REG-020's CAT download/edit/re-upload workflow is descoped; the underlying behavior it verifies (invalid results removed row-by-row, with a reported count in a console warning and a visible note) is covered.
Descoped — secondary decoration (1 row): - SSP-REG-007 (marginal box-and-whisker markers with N / percentiles / mean / SD tooltips) is a secondary axis decoration from the original Webcharts renderer; the core shift comparison (scatter, identity line, and the per-point tooltip of SSP-REG-006) is fully implemented. A follow-up can add marginal box plots if a study needs them.
Adapted — layout (1 row): - SSP-REG-017 ("the listings appear to the right of the chart") is superseded by the shared renderer shell (#17), which standardizes the listing below the chart across every renderer (matching the histogram). Controls-left / chart-above-listing (SSP-REG-016, SSP-REG-018) hold and are enforced by tests/e2e/site.spec.js. SSP-REG-015 ("fits on one page") — flagged vague in the AI review — is addressed by the fixed 460px chart area and the responsive shell rather than a pixel assertion.

Railed participant profile (#99, PPRF-SSP)

The shared participant-profile module mounts in the rail beside the chart (config-on, profile: true) and is fed by the renderer's brush selection via the participantsSelected dispatch — the rail SUPPLEMENTS the linked listing (records vs story), replacing nothing. shift-plot is the rollout's stepper renderer: a brush routinely catches several participants, so the rail collapses to the worst-first cohort stepper ("1 of N"), stepper navigation border-emphasizes the stepped point on the chart without re-dispatching, and a single-point brush shows the full profile directly. The adoption moved the SSP-API-003 dispatch target from the host element to the shell root (PPRF-SSP-004) — backward-compatible because the event bubbles. Adoption rows are PPRF-SSP-001..004 in the participant-profile matrix; unit evidence lives in tests/unit/shift-plot/profile-adoption.test.js.

View-selector rollout (VIEW-3, #76)

One primary display — the baseline-vs-comparison scatter. The linked listing is a brush-selection drill-down shown in addition to the chart, not a mutually-exclusive view switch. Ruled single-view in the shared view-selector rollout (#76) — no view control is added; see view-selector-inventory.md.

Reproducing this report

The evidence set is regenerated from a full test run and committed with the code it qualifies; CI fails when they drift. To verify or rebuild it:

npm ci
npm run evidence:check   # compare a fresh run against the committed evidence
npm run evidence         # regenerate docs/evidence/shift-plot/evidence.json

Screenshot baselines are canonical to the Linux CI runner; the repository's Update evidence baselines workflow is the authoritative way to refresh them. See CONTRIBUTING.md for the traceability convention.