Logistics Research Seminar: Three talks


Speakers


Abstract

Leveraging Transaction Data for Online Pricing and Sourcing of Carrier Capacity in the Truckload Spot Market: Models and Application - Hani Mahmassani, Northwestern University 

Moving goods generates large volumes of transaction data—some of it is available in real-time to support order fulfillment and operational decisions, and when properly retained creates large databases of historical transactions.  These can form the basis of valuable business intelligence and predictive tools to support strategic and operational decisions.  We discuss the use of transaction data as part of a decision support system for online pricing and sourcing capacity in the truckload spot market by third party logistics (3PL) providers.  The transportation spot market consists of shipments handled on a one time load-by-load basis, and exists to facilitate urgent or unfulfilled demand. It is characterized by price volatility and uncertainty in the availability of capacity.

The online decision support system recommends to brokers prices to quote a shipper in real time, along with a list of potential carriers to contact to source the load. At the core of the system are discrete choice models of shipper and carrier acceptance, and an expected profit maximization model. The discrete choice models predict the acceptance or rejection of an offer for a shipment to shippers and a bid for capacity to carriers. The profit maximization model determines the shipper price that maximizes the 3PL provider's expected profit. In addition to these models are procedures for determining and ranking a list of potential carriers for an incoming shipment. The system is applied to real-world data for a 3PL provider, with excellent results.

In addition, we examine in more detail aspects of carrier behaviour underlying their acceptance of shipper loads tendered on the spot market, particularly when multiple loads are bundled together. Carriers’ responses to a hypothetical field experiment are used to estimate carrier reservation prices for bundled shipments, and analysed to assess the effect of bundling on the 3PL’s ability to secure capacity for those shipments more quickly and at better rates. The results indicate that reservation prices can vary considerably across carriers in the same lane. Accounting for the behavioural dynamics of carriers in operational and revenue management strategies can lead to better decision-making, particularly for shippers and their representatives.

Dr. Hani S. Mahmassani holds the William A. Patterson Distinguished Chair in Transportation at Northwestern University, where he is the Director of the Northwestern University Transportation Center.  He has over 30 years of professional, academic and research experience in the areas of intelligent transportation systems, freight and logistics systems, multimodal systems modeling and optimization, traffic science, demand forecasting and travel behaviour, and real-time operation of transportation and distribution systems.  He has served as principal investigator on over 140 funded research projects sponsored by international, national, state, and metropolitan agencies and private industry.  He has published over 270 refereed articles and 140 technical reports.  Past editor-in-chief and current associate editor of Transportation Science, Mahmassani holds current and past editorial responsibilities for Transportation Research C: Emerging Technologies and of IEEE Transactions on Intelligent Transportation Systems.  He has served in an advisory capacity to various institutes and programs, and has performed several program assessments of leading international research institutes and corporate R&D departments. He is emeritus member of Transportation Research Board committees on travel behaviour analysis, telecommunications and travel behaviour, and network modeling.  Mahmassani received his PhD from the Massachusetts Institute of Technology in transportation systems and his MS in transportation engineering from Purdue University.

Efficient Statistical Data Editing - Ton de Waal, Centraal Bureau voor Statistiek & University of Tilburg

National Statistical Institutes (NSIs) and other official statistical institutes have the task to provide high quality statistical information on many aspects of society, as up-to-date and as accurately as possible. This task has to be carried out as efficiently as possible, in terms of budget and response burden. One of the difficulties in performing this task arises from the fact that the data sources that are used for the production of statistical output – traditional surveys, administrative data and Big Data – inevitably contain errors that may affect the estimates of publication figures. In order to prevent substantial bias and inconsistencies in publication figures, NSIs therefore carry out an extensive process of checking the collected data and correcting them if necessary. This process of improving the data quality by detecting and correcting errors encompasses a variety of procedures, both manual and automatic, that are referred to as statistical data editing. In this presentation I will discuss statistical data editing techniques developed for survey data, and examine to which extent they are applicable to Big Data.

Consistency and semantics of big urban traffic data - Henk van Zuylen, Delft University of Technology

Data from urban traffic are becoming more available these days. Data sources are, for instance, probe vehicles with GPS, loop detector data of traffic control systems, Bluetooth scanners, Automated Number Plate Recognition cameras, mobile phone data. Some of these data are owned by commercial institutes that are often not willing to make them available to researchers, others are owned by public institutes but cannot be made available due to privacy  or security reasons. Apart from these commercial ownership and privacy and security issues, it is important that the data that are available for research purposes are consistent and well interpreted.

Traffic data from loop detectors, probe vehicles and ANPR cameras were available for a few days from a medium sized Chinese city (Changsha with 5 million  inhabitants), without limitations of security and privacy. About 26 million records of traffic data are collected every day. Some years ago also data from Bluetooth scanners were  collected in the same city together with data from loop detectors, probe vehicles and visual observations. An analysis was done of the consistency of different observation techniques. Due to technical reasons, travel times and volumes measured from different data sources appear to have various different meanings. Data has to be cleaned and made consistent before they  can be used for further application.