TEST SA
Automate the cleaning of mobility survey data from a research institute
-80%of processing time per study
> 95%of inconsistencies corrected
The need
- TEST is a bespoke research institute, with mobility being one of its areas of expertise. Its surveys reconstruct travellers' real journeys to help public transport networks size their lines and frequencies. Each survey produces between 20 000 and 400,000 lines, with inaccuracies inherent to the field: stops spelled in various ways, incomplete reference data, duplicates. Making them reliable required 10 to 30 days per study.
Galadrim's solution
- Galadrim developed an internal platform dedicated to this step. Teams upload survey files to it: the platform identifies inconsistencies, then automatically corrects them by reconciling labels despite input variations, geolocating stops from public reference data, and reconstructing missing journey points. The corrected file returns in the analysts' working format, with each correction being annotated to remain verifiable.
