Transportation Marketplace Rate Forecast Using Signature Transform

Haotian Gu (), Xin Guo (), Timothy L. Jacobs (), Philip Kaminsky () and Xinyu Li ()
Additional contact information
Haotian Gu: University of California Berkeley, Berkeley, California 94710
Xin Guo: University of California Berkeley, Berkeley, California 94710; and Worldwide Operations Research Science, Amazon.com Inc., Bellevue, Washington 98004
Timothy L. Jacobs: Worldwide Operations Research Science, Amazon.com Inc., Bellevue, Washington 98004
Philip Kaminsky: University of California Berkeley, Berkeley, California 94710; and Worldwide Operations Research Science, Amazon.com Inc., Bellevue, Washington 98004
Xinyu Li: University of California Berkeley, Berkeley, California 94710

Interfaces, 2025, vol. 55, issue 5, 424-436

Abstract: Freight transportation marketplace rates are typically challenging to forecast accurately. However, these forecasted rates impact a company’s overall costs and play a key role in the management of the company’s shipping network and the procurement of supply on the open marketplace. In this work, we develop a novel statistical technique based on signature transforms and build a predictive and adaptive model to forecast these marketplace rates. Our technique is based on two key elements of the signature transform: its universal nonlinearity property, which linearizes the feature space and, hence, translates the forecasting problem into a linear regression, and the signature kernel, which allows for comparing efficiently similarities between time series data. Combined, it allows for efficient feature generation and precise identification of seasonality and regime switching in the forecasting process. An algorithm based on our technique has been deployed by Amazon middle-mile trucking operations with far superior forecast accuracy and better interpretability versus commercially available industry models even during periods of extreme disruption, such as the COVID-19 pandemic beginning in 2020 and the onset of the Ukraine conflict in early 2022. Furthermore, our technique captures the influence of business cycles and the heterogeneity of the marketplace, improving prediction accuracy by more than fivefold with an estimated annualized savings of more than $50 million per year since 2021.

Keywords: Time series forecast; signature transform; signature kernel; two-step LASSO; adaptive regression; non-stationarity; non-linearity (search for similar items in EconPapers)
Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
http://dx.doi.org/10.1287/inte.2025.0251 (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:inm:orinte:v:55:y:2025:i:5:p:424-436

Access Statistics for this article

More articles in Interfaces from INFORMS Contact information at EDIRC.
Bibliographic data for series maintained by Chris Asher ().