Investigating the validity of Aggregated Mobile Phone Data to enhance transport demand knowledge: an evaluation framework

The many changes in mobility behavior observed in recent years require the acquisition of increasingly numerous and accurate data to feed complex forecasting models. Aggregated Mobile Phone Data (AMPD), which provides a large sample and wide temporal coverage, seems promising for analyzing mobility patterns, but its quality can vary. This paper therefore proposes a dedicated, thorough evaluation framework of AMPD, by evaluating its stability and by comparing it with two sources of so-called ground truth data, land use and traffic count data, methods tailored to account for the characteristics of these datasets, borrowed from several fields. The framework was designed to be reproducible elsewhere to assess any similar dataset. As an application case of this framework, we used a dataset provided by Orange in France. Our results in this case provide a validity evaluation of the dataset, encouraging its use for analyzing transport demand volumes.