Self-Census self-report lane The bridge from the citizen app
This is where the two products meet. Citizens self-register in the Self-Census app; an enumerator validates each household through the audited boundary (a minimised, PII-free work list — no names, no NIN); the validated record then joins SpatialIQ as an area-level lane — never per-person, and never overwriting field data.
What needs attention today Priorities
All clear — no critical alerts, no pending decisions.
Population insights Estimation
Building intelligence Classification
Field anomaly flags Discrepancy
Field-capture quality flags (missing structures, possible duplicates, out-of-boundary) — FR-VAL-020. Distinct from the calibration discrepancies (model-vs-truth, FR-VAL-010). Self-Census self-report discrepancies live in the Field Enumerator App lane, not here.
Built-form profile Classification
The mix of building forms — an area-level socio-economic signal. Each bar is a share of classified structures, not people. Classify buildings, count areas.
Model calibration — validation against ground truth Validation
Field-inaccessible areas (0) capture-gap EAs · estimate-only
0 EAs are genuinely field-inaccessible (security / riverine / hard-to-reach). The rest are capture-gap (buildings detected, roster not yet captured). Both carry a modelled range with confidence and method — no field count, no implied validation.