Dockless electric scooters (e-scooter) have emerged as a green alternative to automobiles and a solution to the first- and last-mile problems. Demand anticipation, or being able to accurately predict spatiotemporal demand of e-scooter usage, is one supply–demand balancing strategy. In this paper, we present a dockless e-scooter demand prediction model based on a fully convolutional network (FCN) coupled with a masking process and a weighted loss function, namely, masked FCN (or MFCN). The MFCN model handles the sparse e-scooter usage data with its masking process and weighted loss function. The model is trained with highly correlated features through our feature selection process. Next-hour and next 24-h prediction schemes have been tested for both pick-up and drop-off demands. Overall, the proposed MFCN outperforms other baseline models including a naïve forecasting, linear regression, and convolutional long short-term memory networks with mean absolute errors of 0.0434 and 0.0464 for the next-hour pick-up and drop-off demand prediction, respectively, and the errors of 0.0491 and 0.0501 for the next 24-h pick-up and drop-off demand prediction, respectively. The developed MFCN expands the collection of deep learning techniques that can be applied in the transportation domain, especially spatiotemporal demand prediction.

This study demonstrates the use of mobile phone data to derive country-wide mobility patterns. We identified significant locations of users such as home, work, and other based on a combined measure of frequency, duration, time, and day of mobile phone interactions. Consecutive mobile phone records of users are used to identify stay and pass-by locations. A stay location is where users spend a significant amount of their time measured through their mobile phone usage. Trips are constructed for each user between two consecutive stay locations in a day and then categorized by purpose and time of the day. Three measures of entropy are used to further understand the regularity of user's spatiotemporal mobility patterns. The results show that user's in a high entropy cluster has high percentage of non-home based trips (77%), and user's in a low entropy cluster has high percentage of commuting trips (49%), indicating high regularity. A set of doubly constrained trip distribution models is estimated. To measure travel cost, the concept of a centroid point that assumes the origins and destinations of all trips are concentrated at an arbitrary location such as the centroid of a zone is replaced by multiple origins and destinations represented by cell tower locations. Note that a cell tower location can only be used as trips origin/destination location when a stay is detected. The travel cost measured between cell tower locations has resulted in shorter trip distances and the model estimation shows less sensitivity to the distance-decay effect.

This study applies passive mobile positioning data such as Call Volume, Handover, and Erlang to detect the spatiotemporal distributions of urban activities. We obtained hourly aggregated cellphone data from a dataset of communications in Lisbon, Portugal. Fuzzy c-mean clustering algorithm was applied to the cellphone data to create clusters of locations with similar features in two aspects of activities: the pattern and intensity of urban activities along the hours of a day. In order to validate those clusters as actual predictors of human activity, we compared them with clusters formed using ground truth variables namely presence of people, buildings, points of interest, and bus and taxi movement. To identify the patterns of urban activities, the Erlang data provided a better match, with the ground truth giving 69% of overall accurate predictions. In the case of the intensity of activities, Handover data provided the highest match, with the ground truth yielding 80% of overall accuracy of predictions. Hence, the results demonstrate the potential of passive mobile positioning data in detecting intensity of activities that are superimposed on the different activity patterns, which is a fundamental piece of information for transportation and urban planning.

The development of a trip distribution model requires a massive data collection task, such as expensive travel surveys to identify trip makers' origin and destination zones. Intra-zonal trips are usually ignored in the development of trip distribution models because of the difficulty associated with measuring travel costs. However, ignoring intra-zonal trips leads to incomplete model estimation especially when the zone size is large, and thus, the number of intra-zonal trips is substantial. This is especially important, with the growing interest around the world on creating more walkable and bikeable cities where a high share of those trips is intra-zonal. In this paper, we use mobile phone data to derive country-wide mobility trends. A set of doubly constrained trip distribution models that integrates intra-zonal trips was estimated for 123-district-level traffic analysis zones. We present two approaches to estimate the intra-zonal travel costs measured in terms of trip distance. Our analysis reveals that the average intra-zonal trip distances obtained from the two approaches provide different levels of sensitivity to the distance-decay effect. Our result also shows that model estimation in the absence of intra-zonal trips produces biased estimates.

The movement of trucks represents a significant portion of travel. Surveys have traditionally been used to measure truck movement, but this costly and limited data collection method typically involves in-person interviews and requiring high workload. This study explores different ways in which passive truck GPS data can be used to complement traditional data collection methods, for obtaining detailed information about the travel behaviors of freight trucks. First, we develop a heuristic-based model to identify truck stops. A new methodology is proposed to classify truck stops into a primary or secondary sto, which make identifying trip purposes possible. Primary stops are defined as the locations where the loading or unloading of the goods takes place. Secondary stops are those associated with all other purposes, including refueling, and driver breaks. Finally, we develop a destination choice model for modeling truck movements. This model applies a discrete choice modeling technique to distribute truck trips within the Calgary region, Canada. We test the utility function in the destination choice model to include of business establishment data, travel impedances, and other dummy variables that are likely to influence truck demand. The results show that a combination of trucks’ dwelling times and their entropy can be used to classify truck stops by purpose. This study also shows the potential of using passive GPS data to gain additional insights into truck movements characterization and truck trip distribution modeling.