Training a machine to automate spot pricing of logistics services in a large-scale network

• US 10,332,032 B2
• Filed: 11/01/2016
• Issued: 06/25/2019
• Est. Priority Date: 11/01/2016
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Chowdhary, Pawan R., Ettl, Markus R., Lederman, Roger D., Nonner, Tim, Schimpel, Ulrich B., Xue, Zhengliang, Yang, Hongxia
• Primary Examiner(s): Zhen, Li B.
• Assistant Examiner(s): Lee, Tsu-Chang
• Application Number:
  US15/340,548
• Publication Number:
  US 20180121829A1
• Time in Patent Office: 966 Days
• Field of Search: 706 12
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06Q 10/08 Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising

Abstract

A machine learning algorithm is trained to learn to cluster a plurality of original-destination routes in a network for transporting cargo into a plurality of clusters based on similarities of the original-destination routes, and to learn to cluster the plurality of clusters into a plurality of subgroups based on customer behavior. Influencing criteria associated with each of the subgroups may be determined and based on the influencing criteria, a price elasticity curve for each of the subgroups may be generated. Based on the price elasticity curve and current network traffic, cargo transportation price associated with each of the subgroups may be determined.

Claims

1. A method of training a machine to automate spot pricing of logistics services, the method performed by at least one hardware processor, comprising:

receiving a plurality of original-destination routes in a network for transporting cargo;

clustering the plurality of original-destination routes into a plurality of clusters based on similarities of the original-destination routes;

clustering the plurality of clusters into a plurality of subgroups based on customer related factors,

wherein the clustering is performed by training the machine to learn and store in a memory device a computer-implemented decision tree model and a computer-implemented regression model, and merging execution of the learned decision tree model and the learned regression model to create an interplay between segmentation and regression,

the regression model comprising a logit function modeled as a function of a first parameter representing common effects on cargo attributes multiplied by a vector of attributes for quote k, a second parameter representing origin-destination pair effect multiplied by an indicator value mapping k to the origin-destination pair, and a random noise term;

determining influencing factors associated with each of the subgroups;

based on the influencing factors, generating a price elasticity curve for each of the subgroups;

based on the price elasticity curve and current network traffic, determining cargo transportation price associated with each of the subgroups,

wherein the influencing factors comprise at least rate, chargeable weight (CWT), density, weight, volume, leadtime, number of pieces, freighter route, number of stops, customer industry, product code, special handling flag, seasonality indices indicating CWT demand relative to average CWT, revenue demand relative to average revenue and orders demand relative to average orders, shipper capacity, average price quoted to customer in market, average price quoted to customer overall, customer percent (%) revenue through spot market, and customer % CWT through spot market; and

dynamically balancing load in air traffic network in real-time via providing the cargo transportation price.

2. The method of claim 1, wherein the cargo transportation price is determined for each of the subgroups simultaneously.

3. The method of claim 1, wherein the influencing factors are correlated.

4. The method of claim 1, wherein the determining influencing factors comprises executing a LASSO technique.

5. The method of claim 1, wherein the determining cargo transportation price comprises executing an optimization function that maximizes revenue and a probability that a quoted price will be accepted.

6. The method of claim 1, further comprising visually displaying the subgroups in a graphical user interface.

7. The method of claim 1, further comprising iterating the determining of the influencing factors, generating of the price elasticity curve, and the determining of the cargo transportation price.

8. A computer readable storage medium storing a program of instructions executable by a machine to perform a method of training a machine to automate spot pricing of logistics services, the method comprising:

receiving a plurality of original-destination routes in a network for transporting cargo;

clustering the plurality of original-destination routes into a plurality of clusters based on similarities of the original-destination routes;

clustering the plurality of clusters into a plurality of subgroups based on customer related factors,

wherein the clustering is performed by training the machine to learn and store in a memory device a computer-implemented decision tree model and a computer-implemented regression model, and merging execution of the learned decision tree model and the learned regression model to create an interplay between segmentation and regression,

the regression model comprising a logit function modeled as a function of a first parameter representing common effects on cargo attributes multiplied by a vector of attributes for quote k, a second parameter representing origin-destination pair effect multiplied by an indicator value mapping k to the origin-destination pair, and a random noise term;

determining influencing factors associated with each of the subgroups;

based on the influencing factors, generating a price elasticity curve for each of the subgroups;

based on the price elasticity curve and current network traffic, determining cargo transportation price associated with each of the subgroups,

wherein the influencing factors comprise at least rate, chargeable weight (CWT), density, weight, volume, leadtime, number of pieces, freighter route, number of stops, customer industry, product code, special handling flag, seasonality indices indicating CWT demand relative to average CWT, revenue demand relative to average revenue and orders demand relative to average orders, shipper capacity, average price quoted to customer in market, average price quoted to customer overall, customer percent (%) revenue through spot market, and customer % CWT through spot market; and

dynamically balancing load in air traffic network in real-time via providing the cargo transportation price.

9. The computer readable storage medium of claim 8, wherein the cargo transportation price is determined for each of the subgroups simultaneously.

10. The computer readable storage medium of claim 8, wherein the influencing factors are correlated.

11. The computer readable storage medium of claim 8, wherein the determining influencing factors comprises executing a LASSO technique.

12. The computer readable storage medium of claim 8, wherein the determining cargo transportation price comprises executing an optimization function that maximizes revenue and a probability that a quoted price will be accepted.

13. The computer readable storage medium of claim 8, further comprising visually displaying the subgroups in a graphical user interface.

14. The computer readable storage medium of claim 8, further comprising iterating the determining of the influencing factors, generating of the price elasticity curve, and the determining of the cargo transportation price.

15. A system of training a machine to automate spot pricing of logistics services comprising:

at least one hardware processor;

a display device coupled to the hardware processor;

the hardware processor operable to receive a plurality of original-destination routes in a network for transporting cargo;

the hardware processor further operable to learn to train itself to cluster the plurality of original-destination routes into a plurality of clusters based on similarities of the original-destination routes;

the hardware processor further operable to learn to train itself to cluster the plurality of clusters into a plurality of subgroups based on customer related factors, wherein the hardware processor performs clustering by training the hardware processor to learn a computer-implemented decision tree model and a computer-implemented regression model, and merging execution of the learned decision tree model and the learned regression model to create an interplay between segmentation and regression,

the regression model comprising a logit function modeled as a function of a first parameter representing common effects on cargo attributes multiplied by a vector of attributes for quote k, a second parameter representing origin-destination pair effect multiplied by an indicator value mapping k to the origin-destination pair, and a random noise term;

the hardware processor further operable to determine influencing factors associated with each of the subgroups, and based on the influencing factors, generate a price elasticity curve for each of the subgroups;

the hardware processor further operable to, based on the price elasticity curve and current network traffic, determine cargo transportation price associated with each of the subgroups;

the hardware processor further operable to visually display the subgroups in a graphical user interface on the display device,

wherein the influencing factors comprise at least rate, chargeable weight (CWT), density, weight, volume, leadtime, number of pieces, freighter route, number of stops, customer industry, product code, special handling flag, seasonality indices indicating CWT demand relative to average CWT, revenue demand relative to average revenue and orders demand relative to average orders, shipper capacity, average price quoted to customer in market, average price quoted to customer overall, customer percent (%) revenue through spot market, and customer % CWT through spot market; and

the hardware processor further operable to control dynamically balancing of load in air traffic network in real-time via providing the cargo transportation price.

16. The system of claim 15, wherein the cargo transportation price is determined for each of the subgroups simultaneously.

17. The system of claim 15, wherein the hardware processor is operable to execute an optimization function that maximizes revenue and a probability that a quoted price will be accepted.