A Novel Approach to Reveal Travel Patterns of City Residents Based on Mobile Internet Locating Data

Sensing and mobile internet technologies allow residents’ locational information to be acquired real-time. We propose a novel approach to reveal travel patterns of city residents based on mobile internet location information. We use massive locational data to rebuild travelers’ trajectories among traffic analysis zones (TAZs), extract trip frequencies, travel modes, distance, duration of each trip by staying behavior analysis, trajectory partition, and point of interest (POI) analysis. We calculate travelers’ staying time in each TAZ and get a matrix using 1-h intervals. Using the nonnegative matrix factorization (NMF) method to decompose this matrix, we distinguish different traveler groups and trip types, compensating for missing information using the average in the TAZ to get a dataset similar to resident trip surveys. Results show that our approach can efficiently get travel patterns consistent with traditional resident trip surveys, which can be used for traffic analysis, forecasting, planning, and management.