This paper proposes determining the fraction of trips made by the population using a given mode (private car, bus, metro, bicycle, or walk) in an urban environment by analyzing their historical locations, collected over an extended period of time, at high frequency ('1 min), through their GPS-enabled cellphones. The analysis consists of inferring the mode of transport used by each user at every single location, based on the type of road used, speed, and the distance traveled. Map matching, trip reconstruction, and stay-point determination are essential tools within this method. The proposed method was tuned up by comparing the results obtained with this method versus those obtained in a controlled monitoring campaign over a sample of 52 inhabitants, in which each transport mode was registered manually, and the second-by-second location was registered during a month. Then, the tuned method was applied for the case of the Monterrey Metropolitan area, on a sample of ~0.6 million inhabitants (~12% of the total population), monitoring their location for 3 months at high frequency ('1 min). It was found that in this city, the population prefers private cars (60%) followed by public services (buses and metro) (30%), which agrees with a study based on surveys on the same population in the same year. The proposed method offers a passive, non-invasive, low-cost alternative to determine the distribution of transport modes used by the population with high spatial and temporal resolution, which is key information in the design of public policies oriented to improve mobility and reduce the emissions of greenhouse gases.
