How can mobile phone call data improve transportation demand forecasts?

A recent study showed that anonymized mobile phone call detail records (CDR) can be integrated into the traditional four‑step travel demand model to better predict urban trips. Researchers identified significant stay locations to reconstruct individual mobility patterns, then aggregated them at the zonal level and validated them against a large‑scale household survey conducted in Sri Lanka. The results reveal a strong correlation between CDR‑based estimates and conventional survey outcomes. This approach enables the extraction of origin‑destination matrices and supports route assignment, offering a more accurate, cost‑effective and scalable alternative to traditional travel surveys for urban mobility planning.

Integrating CDR data into the four‑step model marks an advancement in data‑driven mobility. The stay‑based approach identifies prolonged stops from sequences of cellular towers, serving as proxies for activities such as work, school or leisure. These stops are linked to form individual trip chains, which are then expanded to the zonal level using demographic expansion factors. Validation against the Sri Lankan household survey, encompassing over 20 000 families, yielded a Pearson correlation index above 0.85 for origin‑destination matrices. Beyond accuracy, the main advantage lies in the high temporal resolution (hourly) of CDR, enabling capture of demand variations linked to special events, weather conditions or strikes—difficult to obtain with periodic surveys. Limitations include the inability to capture walking or cycling trips when no cellular interaction occurs, and accuracy dependence on cell tower density. Future work could fuse CDR with GPS or electronic ticketing data to fill these gaps, creating a multimodal data ecosystem for more robust transport planning.