Skip to main content

Investigate: Reports & indicators we can build from the 41-source catalogue

IMPLEMENTATION RULES: Before implementing this plan, read and follow:

Status: Backlog​

Goal: Survey the 41 sources Atlas now ingests, identify the reports and composite indicators they enable end-to-end (catalogue → ingest → mart → frontend), and document the conformed dimensions, crosswalks, and reference seeds that have to land in dbt before those indicators become queryable. Output: a sequenced list of indicator-PLAN candidates the user can pick from, plus a clean inventory of the dimensional plumbing each one depends on.

Last Updated: 2026-05-04 (onboarded bufdir-barnefattigdom — Bufdir Barnefattigdom kommunemonitor API; catalogue is now 41 sources)

Origin: Atlas's catalogue grew from 21 to 38 sources between 2026-04-30 and 2026-05-03 (PLAN-007 phase 2 + the FHI onboarding wave). Seventeen new FHI sources extended Atlas from a Samfunnspuls-replication scope into broader public-health-statistics coverage — population projection, immigrant-background mix, youth wellbeing (Ungdata: QoL / depression / painkillers / confiding-friend / alcohol / cannabis / screen-time × 3), primary-care contacts (KPR), and 5-year suicide. The user asked: "create an investigate file on the potential reports and stats we can create based on the data we have gathered. what relations we need to make them happen." This document is the answer — it does not implement anything; it scopes the surface so a follow-up PLAN-* can land each indicator under PLAN-007's catalogue + frontend.

The 41-source catalogue at a glance​

Tagged by topic (Atlas-domain), with eu_theme (DCAT-AP) and geo. All annual unless noted; all kommune-resolved unless noted.

Demographics (11 sources, eu_theme=SOCI mostly)​

SourceWhat
ssb-07459Population by region × sex × single-year age (1986–2026)
ssb-10826Population by bydel/city-total region × sex × single-year age (2001–2026)
ssb-06083Families by family type — single-parent share is a vulnerability proxy
ssb-06913Population change — births, deaths, migration (1951–2026)
fhi-befolkning (338)FHI population composition (counts)
fhi-befolkningsvekst (185)Year-over-year population growth (2002–2024)
fhi-prognose (171)Population projection to 2030 / 2040 / 2050
fhi-innvandrere (175)Immigrant background by country (LANDBAK)
fhi-innvkat (650)1st-gen / 2nd-gen / combined immigrant categories
fhi-bor-alene (187)Adults living alone (16+, by age band)

Income & poverty (5 sources, eu_theme=SOCI)​

SourceWhat
ssb-08764Children under 18 in low-income households (EU/OECD scales)
ssb-06947Whole-population low-income (EU/OECD scales)
ssb-12944Persistent low-income, 3-year rolling, by age group
ssb-06944Household income median by household type
bufdir-barnefattigdomBufdir Barnefattigdom kommunemonitor — kommune-level child-poverty indicator family (annual API time series; triangulates with ssb-08764 on methodology)

Education (4 sources, eu_theme=EDUC)​

SourceWhat
ssb-09429Educational attainment by kommune × sex × level
fhi-vgs-gjennomforing (360)Upper-secondary completion rate (3-year cohorts)
fhi-mobbing (377)School bullying, 7th + 10th grade (3-year averages)
fhi-trangbodd (794)Overcrowded housing by education

Youth & mental health (11 sources, eu_theme=HEAL or EDUC)​

SourceWhat
fhi-neet (809)Not in Education, Employment, or Training, by parents' education
fhi-livskvalitet (373)Subjective quality of life — Ungdata sample-survey
fhi-depresjon (339)Depressive symptoms — Ungdata sample-survey
fhi-smertestillende (390)At-least-weekly painkiller use — Ungdata sample-survey, marker of chronic pain / psychological distress
fhi-fortrolig-venn (354)Has-a-confiding-friend share — Ungdata sample-survey, protective-direction social-connectedness indicator
fhi-alkohol (332)Alcohol use one or more times — Ungdata sample-survey, risk-direction substance-use indicator
fhi-hasj (363)Cannabis use one or more times — Ungdata sample-survey, risk-direction substance-use indicator (heavier suppression than alkohol due to lower prevalence)
fhi-mediebruk-spill (601)>3h/day TV+gaming — Ungdata sample-survey, risk-direction screen-time
fhi-mediebruk-some (602)>3h/day social-media use — Ungdata sample-survey, strongest mental-health-outcome correlation of the three media tables; large sex gap (girls higher)
fhi-mediebruk-underhold (667)>3h/day streaming/entertainment — Ungdata sample-survey, weakest mental-health-outcome correlation of the three; included for time-budget completeness
fhi-selvmord (344)Suicide deaths, 5-year rolling, smoothed MEIS

Welfare & social (3 sources, eu_theme=SOCI)​

SourceWhat
ssb-12131Social-assistance monthly rates set by each kommune
ssb-12132Welfare-income rules — what counts as income (kommune policy)
ssb-13995Social-assistance cases, payouts, support duration

Health-services (2 sources, eu_theme=HEAL)​

SourceWhat
ssb-12292KOSTRA omsorgstjenester — nursing home + home-care indicators
fhi-kpr-1aar (370)Primary-care contacts by ICPC-2 code group (P-codes for psych, K for cardio, L for muscle, "Skader" for injuries)

NGO supply (3 sources, eu_theme=SOCI)​

SourceWhat
ssb-12063KOSTRA kommunale fritidstilbud + voluntary-association count
redcross-branchesRed Cross HQ + Distrikt + Lokalforening with per-branch activities
frrFrivillig Resource Register (private; per-NGO volunteer roster)

Reference (2 sources, eu_theme=GOVE)​

SourceWhat
ssb-klass-kommunerCanonical kommune-code list (Klass 131)
ssb-klass-fylkerCanonical fylke-code list (Klass 104)

Justice / public safety (1 source, eu_theme=JUST)​

SourceWhat
ssb-crime-tablesSSB Px bundle 08484–09406 — reported + investigated offences (national series + kommune-by-place-of-offence averages for 08487)

Reports & composite indicators we can build​

Each row below is an indicator or report-page that the catalogue currently supports. Each names the sources it joins, the conformed dimensions it needs, and the open methodological questions.

1. Kommune Demographic Profile (per-kommune card)​

Reports: total pop now / projected to 2030+2050; year-over-year growth trajectory; age structure (kids share / working-age / elderly); sex ratio; adults-living-alone share; immigrant-background share + 1st-gen/2nd-gen split + country mix; single-parent-family share.

Sources: fhi-befolkning + fhi-befolkningsvekst + fhi-prognose + fhi-bor-alene + fhi-innvandrere + fhi-innvkat + ssb-06083 + ssb-07459.

Required relations:

  • dim_kommune (canonical 4-digit code) — exists already in ssb-klass-kommuner.
  • crosswalk_geo_to_kommune — handle FHI's mixed-level GEO column (kommune/fylke/bydel/national in one dim).
  • dim_age_band — harmonise SSB's single-year Alder with FHI's min_max bands. Pick a canonical partition (suggested: 0-5, 6-15, 16-19, 20-29, 30-44, 45-66, 67-79, 80+).
  • dim_period — splice annual observed (ssb-07459, fhi-befolkning), 3-year rolling growth (fhi-befolkningsvekst's annual series), and projection (fhi-prognose).

Planned additions:

  • ssb-10826 landed as the bydel-level age/sex denominator for Oslo, Stavanger, Bergen and Trondheim. It doesn't change the kommune-resolved Profile, but it closes the denominator gap for Report #14 (Bydel-Level City Profile) as a sister view that turns 4 city-rows into bydel rows.

2. Child Welfare / Vulnerability Composite​

Reports: a per-kommune composite ranking + per-axis breakdown:

  • Child low-income share (ssb-08764 EU60)
  • Persistent low-income share (ssb-12944, age 0-17 slice)
  • Single-parent-family share (ssb-06083)
  • School bullying rate (fhi-mobbing)
  • Overcrowded-housing share (fhi-trangbodd, age 0-17 slice)

Sources: 5 indicators above.

Required relations:

  • dim_kommune ✓
  • ref_indicator_direction — every indicator's "more is worse / better" sign, so a composite z-score has consistent orientation.
  • crosswalk_alder_band — extract the 0-17 slice cleanly from FHI's overlapping bands.
  • Methodology decision: equal-weight z-score? PCA? Per-domain sub-indices (poverty / housing / school)? This belongs in a separate INVESTIGATE since the choice shapes interpretation.

Planned additions (from INVESTIGATE-new-norwegian-public-sources.md):

  • bufdir-barnevern — adds the formal-intervention axis Atlas currently lacks (the prior research notes barnevern as the canonical kommune-level child-welfare measurement).
  • bufdir-barnefattigdom — onboarded (kommune-level time series from the public monitor API). Still pending for this report: explicit composite wiring + Oslo bydel crosswalk alignment (Q3) before the vulnerability card treats Bufdir as a full peer to ssb-08764 at sub-kommune resolution.
  • husbanken-statistikkbank — adds the housing-policy-response axis (bostøtte recipients, kommunal bolig stock) to complement Atlas's existing fhi-trangbodd symptom side.

3. Youth Outcomes Report​

Reports: per-kommune education + employment + wellbeing card for the 15-24 / 16-19 cohorts:

  • VGS completion rate (fhi-vgs-gjennomforing)
  • NEET rate (fhi-neet)
  • Self-reported quality of life (fhi-livskvalitet)
  • Self-reported depression symptoms (fhi-depresjon)
  • At-least-weekly painkiller use (fhi-smertestillende)
  • Has-a-confiding-friend share (fhi-fortrolig-venn) — protective direction
  • Alcohol use share (fhi-alkohol) — risk direction
  • Cannabis use share (fhi-hasj) — risk direction
  • Social-media >3h/day share (fhi-mediebruk-some) — risk direction, strongest mental-health correlation; pair with fhi-mediebruk-spill (TV+gaming) and fhi-mediebruk-underhold (streaming) for full time-budget
  • School bullying (fhi-mobbing, 7th + 10th grade)

Sources: 12 indicators above.

Required relations:

  • dim_kommune ✓
  • crosswalk_alder_band — youth-band consensus across FHI tables (15-24 in NEET, 7+10 grades in mobbing, "1_6" Ungdata cohort identifier in livskvalitet/depresjon — these are not directly comparable).
  • Methodology decision: do we treat Ungdata's 1_6 as a youth proxy, or only present it as an index from FHI without claiming a precise age band? The README for fhi-livskvalitet already flags this as TODO — confirming with FHI's Ungdata methodology is a precondition.
  • crosswalk_kjonn — combined / male / female across all sources.

Planned additions:

  • udir-elevundersokelsen — annual, per-school mobbing + støtte hjemmefra (much sharper than FHI's 3-year-rolling 7th + 10th-grade aggregate).
  • udir-fravar — median absence days/hours for 10th grade + VGS; strong dropout-prediction signal.
  • udir-sluttet-vgs — annual VGS dropout share; complements fhi-vgs-gjennomforing's 3-year completion measure with the within-school-year dropout angle that directorate grants cite specifically.
  • udir-nasjonale-prover — the only direct learning-outcome signal Atlas would carry.

4. Mental-Health Triangulation​

Reports: cross-validate self-report vs care-seeking vs mortality, per kommune:

  • Self-report risk axes (Ungdata): fhi-livskvalitet (low score share inverted) + fhi-depresjon + fhi-smertestillende (frequent painkiller use as a somatised-distress proxy)
  • Self-report protective axis (Ungdata): fhi-fortrolig-venn — having a close confiding friend is a known protective factor
  • Substance-use risk axes (Ungdata): fhi-alkohol + fhi-hasj — the two are correlated; their gap (cannabis high but alcohol low, or vice versa) is itself an interesting cluster signal
  • Screen-time risk axes (Ungdata): fhi-mediebruk-some (highest mental-health correlation) + fhi-mediebruk-spill + fhi-mediebruk-underhold (the 3-domain time budget)
  • Care-seeking (KPR P-codes): fhi-kpr-1aar filtered to KODEGRUPPE ∈ {P01_P29, P70_P99}
  • Mortality (smoothed): fhi-selvmord

Sources: 11 above.

Required relations:

  • dim_kommune ✓
  • ref_icpc2_chapter — map ICPC-2 code ranges (P01_P29, etc.) to chapter labels.
  • dim_period — Ungdata is annual, KPR is annual, suicide is 5-year rolling. Join on the most recent overlapping window.
  • Discrepancy detector: a derived indicator flagging kommuner where self-report ≫ care-seeking (under-treatment) or where suicide rate ≫ self-report (sample bias in Ungdata) — this is the analytically interesting output, not just the four columns side by side.

Planned additions:

  • nav-uforetrygd + nav-aap — registry-side care-claiming counterparts to FHI's KPR (which catches contacts) and Ungdata (which catches self-report). Together they let the discrepancy detector triangulate self-report ↔ contact ↔ welfare-claim ↔ mortality on a single canvas.
  • helfo-fastlege — access axis (do GPs even exist in this kommune?) distinct from use (KPR contact rates) and outcome (suicide). Critical for distinguishing "low contact rate because no doctor" from "low contact rate because no need".

5. Income & Welfare Trajectory​

Reports: per-kommune household-economy story:

  • Median household income (ssb-06944) and trend
  • Low-income share, whole-pop (ssb-06947) and child slice (ssb-08764)
  • Persistent low-income (ssb-12944)
  • Social-assistance caseload + payouts (ssb-13995)
  • Kommune's own welfare policy: ssb-12131 rate-setting + ssb-12132 income-counting rules

Sources: 6 above.

Required relations:

  • dim_kommune ✓ (note: ssb-12131/12132/12292/13995 use KOKkommuneregion0000 — a KOSTRA region code that needs mapping to the canonical 4-digit kommune; mostly identical but with KOSTRA-specific aggregates).
  • crosswalk_kostra_region_to_kommune — already partially handled in existing dbt models; verify coverage.
  • crosswalk_household_type — ssb-06944's HusholdType codes 0000..0004; not yet a dim.

Planned additions:

  • husbanken-statistikkbank — bostøtte recipients, kommunal bolig stock, vanskeligstilte, bostedsløse — the policy-response axis Atlas currently has zero coverage on.
  • nav-helt-ledige — monthly registered unemployment (Samfunnspuls cross-check addition); freshest economic-distress signal at kommune level, complementing annual SSB low-income with much shorter lag.
  • ssb-13006 — average duration on social assistance (Samfunnspuls cross-check addition); pairs with ssb-13995's case-count to distinguish transient from chronic dependency. Resolved 2026-05-05 — already in Atlas via ssb-13995. SSB consolidated the duration data into ssb-13995's ContentsCode dimension (KOSsosgjantmnd0000, KOSgjsnittvklo0000, plus per-age-band variants); table 13006 doesn't exist as a separate API endpoint. See INVESTIGATE-new-norwegian-public-sources.md [Q38] for the verification trail.
  • nav-sykefravaer — quarterly absence rate; the labour-market axis missing from Atlas's current household-economy framing. Note: the four NAV families together also stand up the new Report #15 (Labor-Market Distress Composite).

6. Population-Forecast Planner View​

Reports: per-kommune school-cohort, working-age, and elderly trajectories to 2030 / 2040 / 2050. Use case: kommune planners — "how many primary-school places will we need in 2035? How many home-care recipients?"

Sources: fhi-befolkning (observed) + fhi-befolkningsvekst (recent trend) + fhi-prognose (projected).

Required relations:

  • dim_kommune ✓
  • dim_period that distinguishes observed-AAR vs projected-PROGNOSEAAR — joining requires fhi-befolkning.AAR ↔ fhi-prognose.PROGNOSEAAR, not AAR ↔ AAR (the prognose's AAR is the base year). README for fhi-prognose documents this; the dbt model has to enforce it.
  • crosswalk_alder_band to get a clean school-age (6-15), working-age (16-66), elderly (67+) partition from the 31 ALDER bands.

7. NGO Footprint vs Need​

Reports: per-kommune coverage map:

  • Atlas-tagged need indicators (low-income, NEET, adults-living-alone, mental-health composite)
  • vs NGO supply: Red Cross chapters present + their activity mix (redcross-branches)
  • vs kommune's own count of voluntary associations receiving public support (ssb-12063)
  • "Coverage gap": kommuner with high need indicators but no Red Cross local chapter (or thin activity mix)

Sources: 5+ need indicators + redcross-branches + ssb-12063.

Required relations:

  • dim_kommune ✓
  • dim_chapter (Red Cross side) — already in PLAN-002 marts.
  • dim_activity + the 50→22 category mapping — already in PLAN-002 marts.
  • Gap definition: needs an editorial choice — "high need" = top quartile on composite? Bottom decile on fhi-livskvalitet? Whatever the choice, it's a ref_* decision worth documenting.

Planned additions:

  • brreg-frivillighetsregisteret + lottstift-tilskudd — generalises this report from "Røde Kors-only supply" to "all-Norwegian-NGO supply". Frivillighetsregisteret gives every registered NGO with ICNPO category; Lottstift gives every state grant by org number. Together they form the Norwegian NGO sector at organisational + financial resolution. Sequencing dependency: blocked by INVESTIGATE-tag-indicators-sdg-icnpo.md for the ICNPO crosswalk.

8. Integration Outcomes Gradient​

Reports: do educational and health outcomes vary by immigrant background? Per-fylke (kommune samples too small):

  • Educational attainment (ssb-09429) by sex
  • VGS completion (fhi-vgs-gjennomforing) by INNVKAT
  • Primary-care psych contacts (fhi-kpr-1aar)

Sources: 3 above.

Required relations:

  • dim_fylke ✓ (Atlas already handles is_active filter to avoid pre-2020-reform doubling — see existing memory).
  • crosswalk_innvkat — 2/3/23 → 1st-gen / 2nd-gen / combined.
  • crosswalk_landbak — country-of-origin codes (100 ≈ no immigrant background, etc.) — codes are not human-readable; needs a hand-authored seed against FHI's dimension reference.
  • Sensitivity guidance: integration-outcome reports are politically sensitive; downstream pages should follow careful presentation conventions (no per-kommune naming-and-shaming; aggregate at fylke level minimum).

Planned additions — converts this from a snapshot to a longitudinal pipeline:

  • imdi-bosetting — refugee resettlement inflow (which kommune received how many; quota vs actual). Atlas currently has zero inflow signal.
  • imdi-innvandringsgrunn-kjonn — composition axis (work / refugee + family-reunified / family / education / unknown). Atlas's fhi-innvkat only carries 1st-gen / 2nd-gen / combined; the reason for immigration axis is what integration research uses universally to distinguish labour-migrant from refugee outcomes.
  • imdi-landbakgrunn — country-of-origin (faster cycle than FHI's LANDBAK).
  • Together with nav-uforetrygd / nav-aap (long-term economic outcomes) and fhi-vgs-gjennomforing (educational outcomes), these complete an inflow → composition → origin → education → economic-outcome longitudinal pipeline.

9. Care-Services Capacity vs Population​

Reports: per-kommune per-capita care capacity:

  • Nursing-home + home-care indicators (ssb-12292 ContentsCode subset)
  • Normalised by elderly population share (fhi-befolkning + ssb-07459)
  • Compared with projection: "capacity per elderly resident in 2030 if today's services hold"

Sources: ssb-12292 + fhi-befolkning + fhi-prognose.

Required relations:

  • dim_kommune ✓
  • crosswalk_alder_band — pull elderly slice (67+, 80+) consistently.
  • crosswalk_kostra_kommune — ssb-12292 uses KOKkommuneregion0000.
  • dim_period — for the projection-aware variant, see #6.

Planned additions:

  • helfo-fastlege — primary-care coverage layer beneath KOSTRA omsorg's secondary-care view. With monthly cadence (per [Q5] / [Q32] in the new-sources INVESTIGATE), it also moves Report #9 from a once-per-year snapshot to a within-year trend.
  • helsedir-NKI (Tier-3) — provider-side service-quality complement.

10. School-Capacity Forecast​

Reports: kids 6-15 today vs projected, per kommune. Use case: school construction planning.

Sources: fhi-befolkning + fhi-prognose (filtered to school-age bands).

Required relations: same as #6 + dim_kommune. Simplest indicator in this list — could ship first as a Phase-3 demonstration.

Planned additions:

  • udir-gsi — adds the supply side (current school count + enrolment) to complement the demand side (FHI projection of the school-age population). Also unlocks the new Report #16 (School-System Effectiveness) as a per-school view of the same school inventory.

Forward-looking: reports unlocked by planned new sources​

The 10 reports above are what Atlas's current 41 sources support. The companion investigation INVESTIGATE-new-norwegian-public-sources.md (2026-05-04) still proposes additional sources across Bufdir, NAV, Husbanken, Udir, IMDi, Helfo, DSB, Brreg Frivillighetsregisteret + Lottstift — and a Samfunnspuls cross-check that surfaced four additional Tier-1 source families. Tier-2 crime tables from that file are now onboarded as ssb-crime-tables; Report #11 gains real data (kommune slice from 08487 in fact_kommune_indicators, national slices via the companion indicator models). The ssb-13006 candidate from earlier in this file was resolved 2026-05-05 as a phantom — its data is already in Atlas's ssb-13995 ingest. Each report below remains a new analytical surface beyond the base 10 whenever it still needs editorial or modelling work; updates are noted inline.

The maintenance ritual at the bottom of this file says a new report is justified only when a source "brings a genuinely new analytical axis (categories the current 10 don't cover)". Six new reports below clear that bar — three for new themes (JUST, GOVE, HEAL provider-side), two for new geographic resolutions (bydel, per-school), and one for a domain (labor-market vulnerability) that was lurking inside Report #5 but deserves its own scope once NAV's four indicator families land.

11. Crime / Public-Safety Profile​

Reports: per-kommune registered-offence statistics (two-year averages by place of offence) + national administrative series for investigation outcomes. First-ever JUST-themed Atlas source family.

Sources: ssb-crime-tables ingests Px 08484 (national reported offences by offence type), 08487 (reported offences by gjerningssted — kommune and higher aggregates; two-year averages), 09405 (investigated offences × police disposition, national), 09406 (investigated counts + clearance percentage, national). The kommune slice (08487-derived rows with a 4-digit kommune code) is included in fact_kommune_indicators; national Px tables are exposed as indicators__ssb_08484, indicators__ssb_09405, and indicators__ssb_09406 pending a dedicated national fact pattern.

Pairs naturally with: fhi-mobbing (school violence), fhi-alkohol + fhi-hasj (substance use), fhi-neet (youth disengagement), ssb-13995 (welfare receipt as economic-distress correlate).

Required relations:

  • dim_kommune ✓
  • dim_age_band (shared with Reports 1-4, 6, 7, 9, 10)
  • dim_kjonn (shared)
  • crosswalk_lovbrudd_type (new) — SSB's lovbrudd-type taxonomy → Atlas display labels. Hand-authored seed against SSB's classification.

Open questions:

  • Customer-facing language for "anmeldte lovbrudd" — foreground "registered offences" / "victimisation" framing, not headline crime ranking. Same presentation_policy: 'sensitive' flag as Report #8.
  • Suppression: SSB's small-kommune × type intersections hit "0 or 1" → .. pattern. Atlas's existing SSB ingest already handles this.

12. Preparedness & Resilience​

Reports: per-kommune ROS-analyse status × NGO emergency-response supply. First-ever GOVE-themed Atlas report. Use case: the report Røde Kors's Hjelpekorps and Beredskapsvakt divisions actually need — "is this kommune prepared, and is the NGO presence sufficient to backstop?"

Sources: dsb-kommuneundersokelsen (planned via Tier-2 #8) + redcross-branches (existing — filtered to Hjelpekorps + Beredskapsvakt activities) + later ssb-12063 for the broader voluntary-association count. Could extend with NVE skred/flom hazard layers + MET farevarsler statistics in a v2.

Required relations:

  • dim_kommune ✓
  • ref_preparedness_axis (new) — DSB Kommuneundersøkelsen's question schema typically has 6-8 thematic axes (ROS-analyse, øvelsesfrekvens, kompetanse, samordning, etc.). Hand-authored seed.
  • crosswalk_redcross_activity_to_beredskap — filter for activity-codes that count as emergency-response. Already partially in redcross-activities mart.

Open questions:

  • DSB response rate is uneven; capture is_response to distinguish "low score" from "no submission".
  • Composite-or-two-columns? "Is this kommune prepared" (DSB self-report) and "is NGO supply adequate" (Atlas chapter data) measure different things — surface as a 2D scatter rather than a single composite ranking.

13. Primary-Care Access​

Reports: per-kommune GP coverage (% of population on a fastlegeliste), share of "lister uten fast lege", monthly trend. Pairs with fhi-kpr-1aar care-contact rates (which Atlas has) to distinguish access from use.

Sources: helfo-fastlege (planned via Tier-2 #7) + fhi-kpr-1aar (existing) + later helsedir-NKI (Tier-3) for service-quality complement. Provider-side counterpart to Report #4 (Mental-Health Triangulation) and a refinement of Report #9 (Care-Services Capacity) at the primary-care layer.

Required relations:

  • dim_kommune ✓
  • dim_period with monthly support (per [Q5] / [Q32] in the new-sources INVESTIGATE — Helfo Fastlege is monthly).

Open questions:

  • "Lister uten fast lege" vs "ubesatt liste" — pin Helfo's own definitions in the manifest.
  • Whether to surface this as its own report or fold as an axis inside Report #4 + Report #9. Recommendation: own report, because the access dimension (do GPs even exist?) is qualitatively different from quality or use.

14. Bydel-Level City Profile​

Reports: per-bydel demographic profile + per-capita normalisation of every existing bydel-resolved indicator, for Norway's four bydel-organised cities (Oslo 17 bydeler, Stavanger 9, Bergen 8, Trondheim 4). Demonstrates intra-city inequality — the gap between best-off and worst-off bydel within a city is often larger than the gap between rural kommuner.

Sources: ssb-10826 (bydel-level age/sex population denominator — onboarded from the Samfunnspuls cross-check, [Q48] in the new-sources INVESTIGATE) + Atlas's existing bydel-resolved FHI sources (FHI's GEO column already carries 6-digit bydel codes) + bufdir-barnefattigdom (kommune series live; Oslo bydel rows still need a confirmed Klass/crosswalk before this report can join Bufdir at bydel resolution — see Q3 in the new-sources INVESTIGATE).

Required relations:

  • dim_bydel (new) — Klass 103 (Standard for bydelsinndeling). 38 active bydeler across 4 cities + historical entries; respect SSB's (2001-2003) / (-2019) annotations to avoid pre-reform double-counting (same pattern as dim_kommune's is_active filter).
  • crosswalk_bydel_to_kommune (new) — first 4 digits of bydel-code = kommune-code; trivial. Needed so Atlas's existing kommune-keyed marts can join bydel rows under the parent kommune.
  • crosswalk_geo_to_kommune — FHI's mixed-level GEO column already handles bydel for read; this report formalises the bydel slot as first-class.
  • dim_age_band (shared).

Open questions:

  • Coverage scope — bydel-resolved data only exists for the 4 biggest cities. The frontend needs to handle "this report applies to {Oslo, Stavanger, Bergen, Trondheim}" as a first-class scope, not as an exception.
  • Klass 103 versioning — Oslo's bydeler restructured 2003, Bergen 2019, Stavanger added Finnøy + Rennesøy 2020. Pin the active-bydel filter year per refresh.
  • Why this matters strategically — Atlas's current 10 reports are kommune-resolved and Norway has 357 kommuner with most populated kommuner being big cities. A bydel layer turns the 4 largest cities (~2.1M people, ~38% of national population) from 4 kommune-rows into 38 bydel-rows, dramatically lifting analytical resolution where most users actually live.

15. Labor-Market Distress Composite​

Reports: per-kommune labor-market vulnerability composite covering acute (helt ledige) + transitional (AAP / sykefravær) + long-tail (uføretrygd). Per-axis breakdowns, plus a unified "labor-market stress" score. Connects to but is distinct from Report #5 (which frames the household-economy story).

Sources: nav-helt-ledige + nav-aap + nav-sykefravaer + nav-uforetrygd — all four NAV families planned via Tier-1 #2 in the new-sources INVESTIGATE. Could extend with ssb-06947 (low-income, existing) and the NAV pam-stilling-feed (vacancies, deferred).

Required relations:

  • dim_kommune ✓
  • dim_period — must handle monthly cadence properly per [Q5] / [Q32]. NAV publishes uføretrygd / AAP monthly; sykefravær quarterly.
  • dim_age_band — useful for the youth-NEET axis (Report #3 cross-link).
  • ref_indicator_direction — all four NAV indicators are risk-direction (more = worse); document for composite z-score.

Open questions:

  • Composite-vs-per-axis (same as Reports #2, #4, #7). Recommendation: per-axis pages first, composite later in a methodology-explicit follow-up.
  • NAV-internal kommune groupings — verify per indicator (NAV-region or NAV-kontor catchment is sometimes substituted for canonical kommune).
  • Sensitivity — per-kommune uføretrygd shares in small kommuner are politically charged. presentation_policy: 'sensitive' per [Q35] in the new-sources INVESTIGATE.

16. School-System Effectiveness (per-school)​

Reports: per-school view combining capacity (GSI enrolment + pupil-teacher ratio) + engagement (median absence) + climate (Elevundersøkelsen mobbing + støtte hjemmefra) + outcomes (Nasjonale prøver + VGS dropout). Aggregates to kommune. First per-school (sub-kommune) Atlas report.

Sources: udir-gsi + udir-fravar + udir-elevundersokelsen + udir-nasjonale-prover + udir-sluttet-vgs (all planned via Tier-1 #4 + Samfunnspuls cross-check additions in the new-sources INVESTIGATE).

Required relations:

  • dim_school (new) — first-class school dimension keyed by school org number. Per [Q33] in the new-sources INVESTIGATE: dim_school becomes catalogue-level, not source-mart-local.
  • crosswalk_school_to_kommune (new) — registered kommune for v1, catchment is a separate methodology decision per [Q10] in the new-sources INVESTIGATE.
  • dim_period — annual.

Open questions:

  • Resolution choice — ingest at school level (preferred) and let the mart aggregate to kommune, vs aggregate at staging. Recommendation: school-level raw, mart-aggregated downstream.
  • Small-cell suppression — school-level Elevundersøkelsen suppresses items with <5 respondents. Respect verbatim ([Q11] in new-sources INVESTIGATE); no re-derivation across years.
  • School-identifier stability — Norwegian school org numbers occasionally change at mergers. Need a crosswalk_historical_school to link successor + predecessor numbers.
  • Why this matters strategically — same logic as Report #14: Atlas's kommune-resolved school summaries hide intra-kommune variation between rich-area schools and poor-area schools that's often larger than between-kommune variation.

Conformed dimensions Atlas needs​

Below are the dbt-side artefacts each set of indicators depends on. Many partly exist; a few are missing entirely.

DimensionStatusUsed byNotes
dim_kommune✓ exists (PLAN-002 / dim_kommune)AllAlready filtered to is_active to avoid pre-2020-reform row multiplication (per memory).
dim_fylke✓ existsReports 1, 8
dim_periodpartialAll time-seriesToday individual marts handle their own periodicity (P1Y / P3M / 5Y rolling). Atlas needs a canonical period dim that maps each to a comparable axis (e.g. period_start_year, period_kind ∈ {annual, rolling-3, rolling-5}, is_projection).
dim_age_bandmissingReports 1-4, 6, 7, 9, 10The biggest gap. SSB single-year ages, SSB binned bands, FHI overlapping min_max bands, Ungdata 1_6 cohort identifier — none harmonised. Pick a canonical 8-band partition and map every source's bands to it.
dim_kjonntrivialReports 1, 3, 4, 8Three rows — 0/1/2 to all/male/female. One-line seed.
dim_household_typemissingReport 5ssb-06944.HusholdType codes 0000..0004 — needs labels. Already partially in ref_ssb_household_type seed.
dim_indicatorpartialComposites (Reports 2, 7)An indicator-catalogue table: id, label, source, direction (+1/-1), unit, suppression-policy. Atlas already has scattered per-source indicator views; consolidating them gives Phase-4 frontend a uniform schema for indicator cards.
dim_bydelneeded for Report 14Report 14Klass 103 bydel codes for Oslo, Stavanger, Bergen and Trondheim. Filter to active bydeler per year (Oslo restructure 2003, Bergen 2019). ssb-10826 now supplies bydel population facts; the conformed dimension remains to land.
dim_schoolneeded for Report 16Report 16School org number as PK. Lands with udir-gsi ingest per [Q33] in INVESTIGATE-new-norwegian-public-sources.md — first-class catalogue dimension.

Crosswalks Atlas needs​

CrosswalkStatusUsed byNotes
crosswalk_geo_to_kommunepartialAll FHI sourcesMaps FHI's mixed-level GEO column to canonical kommune / fylke / bydel slots. The complication: 0 = national, 2-digit = fylke, 4-digit = kommune, 6-digit = bydel. Existing dbt views likely handle this implicitly; making it a named crosswalk simplifies new sources.
crosswalk_kostra_region_to_kommunepartialReports 5, 9Maps KOKkommuneregion0000 to standard kommune.
crosswalk_alder_bandmissingReports 1-4, 6, 7, 9, 10The big new one. Source bands → canonical 8-band partition. Some bands map cleanly; some require splitting/aggregating.
crosswalk_innvkatmissingReport 82 / 3 / 23 → 1st-gen / 2nd-gen / combined. 3-row seed.
crosswalk_landbakmissingReports 1, 8FHI's 8 country-background codes. Hand-authored against FHI's dimension reference.
crosswalk_icpc2_chaptermissingReports 4, 9ICPC-2 code-range identifier → chapter / topic. Hand-authored; ~10 entries given the table's KODEGRUPPE values.
crosswalk_aldersrelasjon_ungdatamissingReports 3, 4Maps Ungdata's ALDER="1_6" cohort identifier to a meaningful age band — only after methodology verification with FHI.
crosswalk_bydel_to_kommunepartialReport 14ssb-10826 derives parent kommune_nr from the first four digits of bydel-like Region codes in its source mart. A reusable named crosswalk still needs to land for cross-source joins.
crosswalk_school_to_kommuneneeded for Reports 10, 16Reports 10, 16School org number → registered kommune (per [Q10] in INVESTIGATE-new-norwegian-public-sources). v2 may add catchment-vs-registered methodology.
crosswalk_lovbrudd_typeneeded for Report 11Report 11SSB lovbrudd-type codes → Atlas display labels. Hand-authored seed against SSB classification.
crosswalk_kostra_region_kostragroupneeded for Reports 5, 9, 11Reports 5, 9, 11KOSTRA-group codes (EKG01-EKG17 visible in ssb-13138 metadata). Useful for stratified comparisons (small-rural vs big-city kommuner).
crosswalk_imdi_innvandringsgrunnneeded for Report 8Report 8Work / refugee+family-reunified / family / education / unknown — IMDi's 5-category schema. Lands with imdi-innvandringsgrunn-kjonn.

Reference seeds we'd add​

SeedWhatCost
ref_eu_data_theme✓ already in PLAN-007 phase 2.100
ref_age_bandCanonical 8 bands with min/max + labelsmall (~8 rows)
ref_indicator_directionPer-indicator +1 / -1 signsmall (one row per indicator we surface; ~30-40 indicators eventually)
ref_icpc2_chapterICPC-2 chapter lookupsmall (~17 chapters)
ref_landbakFHI LANDBAK codessmall (8 rows + verified labels)
ref_innvkatFHI INNVKAT codestrivial (3 rows)
ref_kjonnSex codestrivial (3 rows)
ref_bydelKlass 103 bydel labels (Oslo/Stavanger/Bergen/Trondheim) — needed for Report 14small (~38 active rows + historical)
ref_lovbrudd_typeSSB crime-type taxonomy — needed for Report 11small (~12 main types)
ref_preparedness_axisDSB Kommuneundersøkelsen question themes — needed for Report 12trivial (~6-8 axes)
ref_imdi_innvandringsgrunnIMDi 5-category immigration-reason codes — needed for Report 8/17trivial (5 rows)
ref_nav_indicator_familyNAV's helt-ledige / AAP / sykefravær / uføretrygd as a coherent labour-market family — needed for Report 15trivial (4 rows + direction sign)

Open questions for decision​

  1. Composite-index methodology. Reports 2, 4, 7 all involve combining multiple indicators into a single ranking or score. Equal-weight z-score is the default; PCA gives data-driven weights but obscures interpretation. Recommendation: ship per-axis pages first (no composite); add composites in a second-wave INVESTIGATE that explicitly chooses methodology with stakeholders.

  2. Suppression handling. FHI suppresses small counts (NULL value). Reports 4 (mental-health triangulation), 8 (integration), 7 (NGO gap) will hit suppression at fine kommune × subgroup slices. Two options: (a) propagate NULL — incomplete cards; (b) fall back to fylke aggregate — less granular but complete. Recommendation: per-indicator policy in ref_indicator_direction with a fallback_geo_level column.

  3. Smoothing default. FHI publishes both raw RATE and smoothed MEIS for small-area indicators (suicide, primary-care contacts). Atlas already chose MEIS for fhi-selvmord and RATE for fhi-kpr-1aar. Should there be a global "use smoothed where the source provides it" preference? Recommendation: yes, default to smoothed; document the choice in dim_indicator.

  4. Period harmonisation. Annual + 3-year rolling + 5-year rolling sources have different time semantics. Joining them naively produces wrong answers (e.g. averaging an annual rate against a 5-year rate). Recommendation: a dim_period with a period_kind discriminator that downstream marts must respect; the frontend exposes this in indicator metadata.

  5. Sample-source caveat ("Ungdata is not a census"). fhi-livskvalitet and fhi-depresjon are sample-survey aggregations. They're not directly comparable to enumeration-based sources like fhi-befolkning. Recommendation: tag indicators with data_quality_kind ∈ {enumeration, register, sample, projection} so the frontend can render appropriate caveats.

  6. Suicide-publication policy. fhi-selvmord's README notes that downstream consumers should follow FHI's veileder for omtale av selvmord. Recommendation: tag the indicator with presentation_policy: 'sensitive' and have the frontend surface a banner / policy-link on those pages.

  7. NGO-gap definition. Report 7 needs an editorial choice on what "high need" means. Recommendation: settle this with NGO partners (e.g. Røde Kors, Folkehjelp) — it's a sector-knowledge call, not a data-modelling call.

Sequencing recommendation​

The smallest indicator that demonstrates the catalogue in /data end-to-end is #10 School-Capacity Forecast — two sources, one period join, one age-band crosswalk. Ship it first as the Phase-3 / Phase-4 demonstrator, then layer:

  1. #10 School-Capacity Forecast — proves the period/age-band plumbing.
  2. #1 Kommune Demographic Profile — adds the immigrant-background slice + bor-alene; proves the multi-dim FHI joins.
  3. #3 Youth Outcomes — first cross-topic (education + health) report; uses Ungdata sources, exposes the sample-vs-enumeration caveat.
  4. #5 Income & Welfare Trajectory — first SSB-only report; tests the KOSTRA-region crosswalk.
  5. #9 Care-Services Capacity vs Population — first projection-aware report.
  6. #4 Mental-Health Triangulation — first cross-validation report; exposes discrepancy as the interesting signal.
  7. #7 NGO Footprint vs Need — last because it depends on most of #1-#5 and adds the NGO-side joins.

Reports 2, 6, 8 are higher-leverage but each opens a methodology question that benefits from earlier reports' learnings.

Forward-looking reports (#11–16) plug in once the corresponding planned sources land per the new-sources INVESTIGATE sequencing. Suggested layering on top of #1–#10:

  1. #10 → #16 — once udir-gsi lands for Report #10, the per-school resolution it pioneers (dim_school) makes Report #16 (School-System Effectiveness) the cheapest follow-up: same source family, same dim, more axes.
  2. #1 → #14 — with ssb-10826 landed, Report #14 (Bydel-Level City Profile) is the cheapest big-impact addition once the conformed dim_bydel / reusable bydel crosswalk work lands.
  3. #5 → #15 — once the four NAV families land for Report #5's expansion, Report #15 (Labor-Market Distress Composite) folds them into a coherent labour-market story (with the same composite-vs-per-axis question as Reports #2, #4).
  4. #11 Crime / Public Safety — SSB Px bundle ssb-crime-tables is ingested (08484, 08487, 09405, 09406). Product work remaining: tighten customer-facing wording (Q18), optional crosswalk_lovbrudd_type, and richer joins with FHI survey registers already in Report #4/#3.
  5. #13 Primary-Care Access — once helfo-fastlege lands, separates access from use in the mental-health and care-services stack.
  6. #12 Preparedness & Resilience — last in the new-report wave because it depends on dsb-kommuneundersokelsen AND on the NGO-side activity filter. The most operationally useful report for Hjelpekorps / Beredskap stakeholders.

What this investigation does NOT cover​

  • Implementation: each row above becomes its own PLAN-* once the user picks one. This file is the menu, not the recipe.
  • Frontend design: how indicator cards / chart compositions / filter sidebars look on /data. That's PLAN-007 phase 4.
  • External integrations: federated discovery via DCAT-AP-NO, MCP exposure, etc. — separately tracked in INVESTIGATE-felles-datakatalog-classification.md, INVESTIGATE-data-discovery-surface.md.
  • Single-NGO supply analytics: covered in PLAN-002 + supply marts; this catalogue augments demand-side tools but doesn't change supply-side modelling.

Maintenance — keep this current as new sources land​

This document is the menu of reports the catalogue currently supports. When a new ingest source lands (the user pastes an FHI / SSB / etc. URL and we onboard it), this file MUST be refreshed in the same commit.

Per-new-source checklist:

  1. Update the source count in the title, the Goal paragraph, the Origin paragraph, the ## The N-source catalogue at a glance heading, and the Cross-references line that names atlas-data/ingest/src/sources/ count.
  2. Slot the new source into the right thematic cluster under "The catalogue at a glance" — Demographics / Income / Education / Youth & mental health / Welfare / Health-services / NGO supply / Reference. Bump the cluster's source count in the heading.
  3. Walk the 10 reports and identify which gain a new column. Common patterns:
    • A new Ungdata-shaped source slots into Report 3 (Youth Outcomes) and Report 4 (Mental-Health Triangulation).
    • A new income / welfare source slots into Report 5.
    • A new demographic source slots into Reports 1 + 6 + 9 + 10.
    • A new health-services source slots into Reports 4 + 9.
  4. Consider whether the new source enables a brand-new report (an 11th, 12th…) — only when the source brings a genuinely new analytical axis (e.g. environmental data, transport flows, crime statistics — categories the current 10 don't cover). Add the new report with the same shape: short reports paragraph, source list, required relations, methodology question. Update the sequencing recommendation to slot it in.
  5. Note any new dimension or crosswalk the source introduces (e.g. new FHI dimension code like KODEGRUPPE, INNVKAT, LANDBAK) in the "Conformed dimensions Atlas needs" / "Crosswalks Atlas needs" tables.
  6. Update Last Updated at the top.

The contributor guide (website/docs/contributors/ingest-modules.md) references this maintenance ritual in its adding-a-source workflow — don't ship a new source without doing the walk-through above.

Cross-references​