ORIGINAL COMPANY-CONTEXT PRACTICE
Estimate synthetic coverage from biased device reports
Original practice scenario: A hypothetical connectivity dashboard uses measurements uploaded by volunteer devices, with many more reports in cities than rural areas.
Written for independent practice using company context; this is not a verified question from a company interview.
A REASONED APPROACH
Model answer.
Crowdsourced measurements describe observed devices and locations, not automatically geographic or population-wide coverage. Define the coverage target before selecting a weighting or interpolation method.
Work through it
- Audit reporting density, device capabilities, time of day and participation patterns; separate absent measurements from measured poor service.
- Evaluate against a deliberately sampled synthetic reference grid, reporting uncertainty for sparse locations and limitations of any spatial interpolation.
- Compare device and area slices, use temporal holdouts and assess sensitivity to participation bias and environmental changes.
Tradeoffs to discuss
- Population weighting answers a different question from geographic-area weighting; both can be useful when their denominators are explicit.
Common pitfalls
- Do not color an unmeasured region as covered by default.
- A dense cluster of repeated measurements should not dominate results simply because one device reports frequently.