This layer is a synthetic but constraint-driven workplace system. It is designed to be operationally realistic for mapping,
assignment, simulation, and inspection, while staying honest about what is modeled versus what is directly observed.
What is exact
- Municipality by industry employment totals are controlled to the official SCB totals used in the build.
- Each workplace carries an explicit profession mix, and the sum of those mixes reproduces the calibrated occupation distributions used for the municipality, county, or national fallback cell.
- Each workplace also carries an archetype derived from the best available local evidence: official site category, customer-facing POI category, name patterns, profession mix, and sector fallback.
- The map now uses a hierarchy of location evidence: exact source points when available, street/address interpolation where defensible, real address-point placement inside DeSO, sector-interleaved remixing for generic DeSO fallback rows, and education-site anchoring for the education sector.
- The viewer switches scale deliberately: 5 km grids at broad zoom, 1 km grids at intermediate zoom, and individual workplace icons at local zoom. This keeps national patterns legible without hiding site-level detail when you zoom in.
- The legend and top switch bar expose the same calibrated industry filter; they change only what is displayed, not the underlying employment totals.
What is modeled
- Named workplace headcounts are synthetic allocations, not payroll-register headcounts for a legal establishment.
- Archetypes are interpretable labels for site function, not claims that every point was matched to a unique audited establishment record.
- The employee table you see after clicking a workplace is an expanded representation of the calibrated profession mix, not an identified person register.
- Where a sector is too aggregate for direct establishment counts, workers are distributed over plausible site anchors rather than left concentrated in centroid clusters; generic fallback rows are also remixed within each DeSO to avoid visible one-sector street blocks before the layer is summarized into grids at small scales.
Source stack
- SCB municipality by industry employment controls provide the hard totals.
- SCB/occupation controls provide the profession distributions.
- HERE address points, streets, and POIs provide spatial realism.
- SCB preschools and education-site anchoring are used to avoid implausible giant school workplaces.
- High-salience retail brands can be promoted from HERE POIs onto plausible calibrated retail rows so customer-expected stores such as IKEA are visible without changing municipality totals.
- Customer-facing shops, cafes, fast-food sites, fuel stations, kiosks, pharmacies, and banks can also be promoted from POI evidence into calibrated `G`, `I`, and `K` rows to improve storefront realism without breaking sector totals.
- Official school, healthcare, and government-site enrichment is used where it improves naming realism without breaking the calibrated totals.
- UKÄ, Socialstyrelsen, Skolverket, Trafikanalys, SCB open geodata, HERE, OSM, 1177, and GTFS are cached locally as the business-register-free backbone for archetype and profession refinement.
- The viewer layer itself is interactive: industry filtering is a display tool applied on top of the calibrated source file, not a separate resynthesis.
Why this is defensible
- The model prioritizes correct totals and plausible spatial dispersion over false establishment-level precision.
- Archetypes make the synthetic layer easier to reason about by separating supermarkets from kiosks, clinics from care homes, universities from schools, and warehouses from taxi operators even when sector codes are broad.
- That tradeoff is intentional: for scenario work, assignment, and map interpretation, a calibrated synthetic site system is more useful than a partial real-world list with missing sectors and missing headcounts.
- Weak records were not hidden; instead, the high-visibility failure modes were progressively corrected: centroid circles were replaced, oversized education sites were rebalanced, and large-site naming was separated from raw employee allocation logic.
Known limitations, stated plainly
- This is not a legal-establishment census and should not be read as one.
- Some institutions appear on multiple campus rows because the source POI layer contains multiple valid campus points.
- Small exact fragments from job or directory sources can still exist alongside larger rebalanced institutional sites; this reflects mixed evidence rather than hidden duplication.
- Site-level headcounts are best interpreted as operational workplace load at mapped anchors, not audited HR totals.
- Medium- and low-confidence archetypes in the synthetic tail remain useful for simulation and map reasoning, but they should not be mistaken for exact legal-establishment classifications.
Best use
- Use this layer for spatial analysis, synthetic agent assignment, profession-aware simulation, workplace inspection, and map communication.
- Use official payroll or establishment registers only if the task requires audited employer-of-record headcounts at site level.
