
Choose geographic granularity for data analysis
Data for a municipality, department or station describes different objects. Choose the scale answering your question before matching maps and files.
Start with the territorial question
Specify the use: locating a service, comparing territories or describing change. Identify the relevant population or objects. The finest scale is not automatically the most useful.
Fictional example: selecting a workshop location also requires access and travel-time information. A departmental average may provide context without answering the local decision.
Retain boundaries and their dates
Record geographic boundary, identifier and vintage. Files may use the same names with different boundaries or zoning. Check these before joining.
Retain cases where a code changes or a zone groups multiple units. Similar labels do not justify automatically assigning an observation to a territory.
Distinguish point, area and population
A station measurement relates to a point and observation conditions. A territorial total relates to an area. A rate adds a reference population. Document this transition before generalising.
Do not turn a point observation into a diagnosis of the whole area. When aggregating observations, explain the rule and gaps in coverage.
Make the map verifiable
Provide title, unit, period, legend and source. Distinguish missing values from zero. Ensure text communicates information without depending solely on colour.
A map may guide a question without explaining causation. Retain a table or export of used data so the comparison can be reviewed.
Compare territories with changed boundaries
Merged municipalities or revised areas can break comparisons between vintages. Keep boundary dates and find an appropriate crosswalk. An unchanged name does not establish unchanged area or population.
Separate additive quantities from rates and averages. Recalculate rates from compatible counts when available. A simple average of percentages across areas does not necessarily represent the rate for their combined population.
Separate missing data from low values on maps
Represent missing observations differently from low values. State exclusions, observation date and actual geographic resolution. Replacing missing values with zero can create a misleading impression of complete coverage.
Check interpretation without colour through legends, labels and a data table. Have responsible specialists review small sensitive counts before publication. Finer resolution increases apparent precision without establishing reliability or permission to publish.
A record to keep with the decision
| Item | Evidence |
|---|---|
| Question | Use, audience and territorial object. |
| Boundary | Scale, identifier and vintage. |
| Measurement | Point/area, unit and population. |
| Presentation | Legend, missing values, source and table. |
Frequently asked questions
Is finer granularity always better?
It may increase apparent precision and uncertainty. Choose for the question, coverage and data limitations.
Can an average be assigned to each resident?
No. Aggregated statistics do not automatically describe every person or location.
Reference material
Observatoire des territoires · Cartographie
The practical checklist is an editorial synthesis to adapt to your service. It does not constitute a certification or an audit result.