Urban Freight Delivery Stop Identification with GPS Data

Abstract

Delivery stop identification is a crucial but challenging step in the measurement of urban freight performance. This paper presents the application of a robust learning method, support vector machine (SVM), in identifying delivery stops with GPS data. The duration of a stop, the distance from a stop to the center of the city, and the distance to a stop's closest major bottleneck were extracted as the three features used in the SVM model. A linear SVM with nested K-fold cross validation proved to be highly reliable and robust in identifying delivery stops with relatively long stop duration, such as those made for grocery stores. Second-by-second freight GPS data collected in New York City were used to conduct a case study. The identification accuracy for the case study was higher than 99% for the training and testing data sets.

Extracting trips from multi-sourced data for mobility pattern analysis: An app-based data example

“…In this section, we present results on stays and trips that are extracted following the DCI framework. The results are compared with the state-of-the-art SVM model that is designed for GPS data (Yang et al, 2014) as well as external data sources to examine the effectiveness of the framework.…”

“…To be robust against thresholds selection, some machine learning methods were proposed to identify stays from GPS data (Gong et al, 2015;Yang et al, 2014). Yang et al (2014) used a Support Vector Machine (SVM) classifier to identify delivery stops (stays) of trucks from freight GPS data. They extracted three features for the SVM model: stop duration, distance between a potential stop location to the city center as well as to the closest major bottleneck (such as bridges and tunnels).…”

“…One challenge of applying machine learning methods to passively generated data is to obtain ground-truth data for training. The study in (Yang et al, 2014), for example, trained the SVM model using delivery logs from truck drivers and some manually labeled training datasets from the authors; Zhou et al (2017) trained their model using data acquired by soliciting users' travel patterns via online surveys.…”

“…Contrast to actively solicited data such as household travel survey data, the passively-generated data are usually location-and time-stamped and are the by-product of some nontransportation related purposes (Chen et al, 2016). Among them two kinds of data are widely used: cellular data generated from cellular networks for mobile phone billing purposes (Alexander et al, 2015;Wang and Chen, 2018) and vehicular GPS data generated from the processes of, for example, freight/taxi/buses management and operation (Yang et al, 2014). Besides data that are generated from single positioning technology (e.g., cellular tower based or GPS), data are also emerging from the combined use of multiple positioning technologies, including GPS-, WiFi-, Bluetooth-, and cellulartowers-based technologies.…”

Integrating mobility data sources to define and quantify a vehicle-level congestion indicator: an application for the city of Turin

“…The methodology presented in the paper is based on the integration of these two datasets. GPS data provided by vehicle fleets of trucks or delivery vans, such as those collected in Dataset2, have usually been exploited to retrieve information on various aspects, such as freight performance measures [20], commercial vehicle tour activity [21], trucks routing behaviours [22] or delivery stops identification [23]. Although some examples of traffic data fusion coming from different sources could be found in literature [24][25][26], one of the main innovative features of the presented methodology lies in the procedure used to integrate a dataset coming from the road infrastructure side with one derived from road users.…”