01661043

Component

This study analyzes the shipment status of 14,858 crowd-shipping requests recorded between January 2015 and December 2016 throughout the U.S. The authors apply the random forest machine learning algorithm to predict bid, acceptance, and delivery status, while using shipping request and built-environment and socioeconomic features as explanatory variables. The results demonstrate that built-environment and socioeconomic variables are useful features in predicting shipment statuses where shipping request and package information is unknown. Contingent upon the results of the 10-fold cross-validation technique, the models show promise in their ability to effectively predict shipment status in real time and to be integrated into smartphone-based crowd-shipping applications. Using only the shipping request and package features, the bid, acceptance, and delivery statuses are predicted with the accuracy of 80.66%, 78.34%, and 98.59%, respectively. Incorporating only built-environment and socioeconomic features, the bid, acceptance, and delivery statuses are predicted with the accuracy of 80.07%, 74.21%, and 70.07%, respectively. In practice, the models presented in this study show promise in their ability to effectively predict shipment status in real time and to be integrated into app-based crowd-shipping systems. They foster many benefits for the crowd-shipping market that reach across the spectrum.

This paper was sponsored by TRB committee ADB40 Standing Committee on Transportation Demand Forecasting. Alternate title: A Dynamic Prediction of Shipment Status in Crowdsourced Delivery Platforms.

18-06436

English

7p

2018

Transportation Research Board 97th Annual Meeting

Digital/other

References; Tables

Crowdsourcing; Delivery service; Mathematical prediction; Shipments

Planning and Forecasting; Transportation (General)

Transportation Research Board Annual Meeting 2018 Paper #18-06436

TRIS, TRB, ATRI

Jan 8 2018 11:40AM