Ideal Test Plan

IDEAL TEST PLAN (Unlimited Time/Resources)

A. UNIT TESTING (Automated - Vitest with React Testing Library)

1.1 Utility Function Tests (src/utils/)

• urlManager.js: URL parameter parsing, encoding, default values

• mapUtils.js: State/location mapping functions

• datasetUtils.js: Dataset configuration retrieval

• forecastleValidation.js: Forecast input validation

• forecastleScoring.js: Scoring calculation accuracy

Test Cases:

• Valid inputs (state codes, city abbreviations, date formats)

• Edge cases (empty strings, null values, malformed data)

• Boundary conditions (min/max values for scales)

• Data transformations (city abbr → state code mapping)

• URL parameter round-trip (encode/decode)

1.2 Hook Tests (src/hooks/)

• useForecastData.js: Data fetching and caching

• useView.js: Context consumer validation

• useQuantileForecastTraces.js: Trace data generation

• useOverviewPlot.js: Plot data aggregation

• useAdaptiveYRange.js: Y-axis range calculations

Test Cases

• Data fetch success/failure scenarios

• Missing or malformed JSON responses

• File not found (404) handling

• Empty datasets (no forecasts)

• Different state/location combinations

• Chart scale transformations (linear/log/sqrt)

• Quantile presence/absence in data

1.3 Context Tests (src/contexts/ViewContext.jsx)

• Location selection persistence

• View type changes with state preservation

• Model/date/target selection independent per view

• Location validation when switching views

• Chart scale/interval/legend settings persist across views

• City abbreviation → state code mapping on view switch

• Invalid location handling (state without MetroCast)

• Success Criteria: State properly persists and validates.

B. COMPONENT TESTS (React Testing Library)

2.1 StateSelector Component Tests

• List renders all available states/cities

• Search filters work (case-insensitive)

• Location selection updates context

• Keyboard navigation (arrow keys, enter)

• No default highlighting when nothing selected (Bug #3 fix)

• MetroCast view shows city hierarchy

• Non-MetroCast view shows flat list

• Mouse and keyboard interaction equity

2.2 ViewSwitchboard Component Tests

• Correct view component renders per viewType

• Data props passed correctly to view components

• Error boundaries catch component failures

• Loading state displays appropriately

• View switching resets/persists state correctly

2.3 Attribution/Citation Component Tests

• Attribution info displays for selected datasets

• Multiple attributions visible simultaneously

• Links to data sources are active

• Attribution not obscured by selection menu

• Citation data structure matches requirements

2.4 Chart Control Components Tests

• Y-scale selection (linear/log/sqrt) applies correctly

• Interval visibility (median/50%/95%) toggles work

• Legend display toggle persists

• Settings apply across multiple charts

• Scale transformations produce correct values

C. INTEGRATION TESTING

3.1 View Switching Workflows

a) Flu MetroCast (city selected) → FluView

• • Expected: City abbr maps to state code (Bug #1 fix)

  • Verify: FluView loads state-level data correctly; if city abbr invalid, shows error then resets (Bug #2 fix).

b) COVID-19 (state NY) → Flu MetroCast

• Expected: NY → Default MetroCast state if not supported.

• Verify: No errors displayed; MetroCast loads correct default.

c) FluView (CO selected) → MetroCast

• Expected: MetroCast loads Colorado data.

• Verify: City list available for Colorado; UI not obscured by selection panel.

d) RSV (VA selected) → MetroCast

• Expected: MetroCast loads Virginia data.

• Verify: VA supported in MetroCast.

e) Return to Previous View

• Expected: Previous selections restored.

• Verify: URL contains view history; State persists without re-fetching.

3.2 Data Integrity Testing

• Load all 13 MetroCast states: Verify no console errors, cities have correct state associations.

• Load all FluSight states: Verify all metadata.json present, targets consistent, model lists complete.

• Test data consistency across views: Verify ground truth data identical across views.

• Empty forecast scenarios: Verify "No data available" message instead of crash.

3.3 State Persistence Testing

• Session storage of URL parameters: Reload restores all selections (?view=flu_forecasts&location=CO&date=2026-03-15).

• Chart settings persistence: Scale/intervals default to user choice and persist across view switches.

• Model/date selection persistence: Selected models and dates stay selected when switching views or navigating back.

• Invalid state recovery: Selecting invalid location → automatic reset.

3.4 Attribution/Citation Workflows

• Display multiple data sources: Attribution visible for Flu + MetroCast; not cluttered.

• Citation export functionality: Generates formatted bibliography matching academic standards.

• Link validation: All links functional, point to official sources, and open in new tabs.

D. SYSTEM/END-TO-END TESTING (Manual - Researcher Workflow)

• 4.1 Explore Flu Forecasts: Select CO -> Denver -> Switch to FluView (Verify CO loads) -> Return to MetroCast (Verify Denver selected).

• 4.2 Comparative Analysis: Load Flu TX -> COVID TX -> RSV TX -> Compare attribution sources.

• 4.3 State Change Workflow: Load MetroCast CO -> Select Denver -> Navigate to RSV -> Automatically switches to CO-level.

• 4.4 Error Handling: Manually construct URL with invalid location -> Attempt unsupported combo -> Clear recovery options shown.