Before data is transformed into metrics, a set of data cleansing, validation, normalization and approximation processes is performed:
- There are basic machine-readable format checks and anomaly detections performed by our automatic data pipelines.
- The data is filtered using satellite and building/road/water footprint data to match with actual territory of interest.
- Data validation checks are run with internal quality reports production.
- Then, data normalization is carried out for making data comparable over time and geography.
For approximation, we apply wide historical samples as well as official Census data.
We guarantee quality of our metrics, and will be glad to share our validation results with you if required.