Efficiency / France / Media & Tech
Large scale mobility synthetic data
"The precision in the analysis of daily trips will allow our advertisers to program their campaign with great accuracy according to their specific audience needs." Chairman, Client
Business issue
Our client, a major media player in out of home media, needed to create a mobility base of France at a very granular level in order to assess Out Of Home media performance.
The objective was to understand, for any given street in France and at any given time in the year, how many people were in the street. They wanted this broken down by which mode of transportation they were using, as well as the sociodemographic travel purpose.
Our solution
To achieve this, we combined open data, 3rd party paid data, survey data, and metering data, and matched this and aggregate traffic with individuals’ trips. From a respondent base of 18,000 “real” individuals that carried advanced meters, we expanded to a synthetic virtual population of 4 million individuals representing the French population. 1.5 billion trips were generated by this population over a year. We then matched aggregate traffic from calibration data with individuals’ trips.
We developed a bespoke, and very scalable, Agent Based Model for the 4 Million synthetic population. These 4 Million virtual individuals were designed to be representative of the French population. All of them had a home, some had a workplace or a school, some went shopping in malls, some went to airports or drove to a secondary home for holidays.
We then processed the data using a cloud data platforms, including Google Cloud Platform (GCP) and Big Query, for faster delivery feeding into a hybrid architecture platform to leverage the synthetic data and deliver the required KPIs in real-time.
For more information on this case study, get in touch here.
Impact
Compared to the former platform, the new one included much more granularity, such as time and space precision, in turn allowing our client to best deliver value to it’s ecosystem.