README.md

metadata

pretty_name: Flu MetroCast Hub — sub-state flu hosp forecast target
license: cc-by-4.0
size_categories:
  - 10K<n<100K
tags:
  - cadence-weekly
  - geo-us
  - surveillance-respiratory
  - pathogen-influenza
  - tier-2
  - availability-open
schema_version: '0.1'
source_id: flu-metrocast-hub
source_url: https://reichlab.io/metrocast-dashboard/
manifest_section: §4.6
surveillance_category: respiratory
pathogens:
  - influenza
availability: open
availability_notes: null
access_type: github
tier: 2
cadence: weekly
geography_levels:
  - subnational-city
geography_countries:
  - US
gold_standard_for: []
vintaged_version_of: null
succeeds: null
derived_from: []
value_columns:
  - name: value
    unit: admissions/week
    value_type: incident
    description: Flu hospital admissions / ED visits for the metro area
    aggregation: sum
notes:
  extra_columns:
    - column: target
      description: Hubverse target identifier (e.g. 'wk inc flu hosp metro').
    - column: location_name
      description: >-
        MetroCast metro identifier (city / county). `location_id` is
        `US-METRO-<id>` synthetic.
  interpretation_caveats:
    - column: value
      caveat: >
        Metro definitions vary by jurisdiction (NYC = 5 boroughs; LA = LA
        County; etc.).

        Don't sum metros to recover state — they don't tile cleanly.
  general: |
    Truth-only ingest, model forecasts handled separately.
extra:
  schema_proposal: >
    `subnational-city` location level needed a synthetic prefix (`US-METRO-`)
    because the

    schema's regex for that level is permissive but un-anchored to a registry.
    Worth

    formalising: maintain a `metro_registry.yaml` with city codes and human
    names.
computed:
  last_ingested: '2026-04-26T12:48:50Z'
  row_count: 132242
  time_coverage:
    - start: '2024-08-03'
      end: '2026-04-18'
  geography_unit_count: 77
  observed_cadence_days: 7
  missing_gaps: []
  data_hash: 9b5355d031f000d2

Flu MetroCast Hub — sub-state flu hosp forecast target

Truth-only ingest, model forecasts handled separately.

Source: https://reichlab.io/metrocast-dashboard/

Coverage

• Time: 2024-08-03 → 2026-04-18
• Cadence: weekly (observed median spacing: 7 days)
• Geography levels: subnational-city — 77 unique location IDs
• Countries: US
• Pathogens: influenza
• Surveillance category: respiratory
• Rows: 132,242

Columns

Column	Unit	value_type	Aggregation	Description
value	admissions/week	incident	sum	Flu hospital admissions / ED visits for the metro area

Additional data columns

• target — Hubverse target identifier (e.g. 'wk inc flu hosp metro').
• location_name — MetroCast metro identifier (city / county). location_id is US-METRO-<id> synthetic.

Interpretation caveats

Things that may differ from how other sources define a similar measure. If you're combining this dataset with another, read these first.

• value — Metro definitions vary by jurisdiction (NYC = 5 boroughs; LA = LA County; etc.). Don't sum metros to recover state — they don't tile cleanly.

Access

• Availability: open
• Access type: github
• License: cc-by-4.0
• Tier: 2

---

Schema version 0.1 · Last ingested 2026-04-26T12:48:50Z · source_id: flu-metrocast-hub · Manifest section §4.6