Systems and methods for imputation of transshipment locations

1. A computer-implemented method for imputation of transshipment locations, the method comprising:

training, by a computing system comprising one or more computing devices, a machine learning model (i) using a set of training data indicating a set of historical outcomes of a set previous shipments, (ii) based on a loss function evaluating a difference between a model prediction and a ground truth label being backpropagated through the machine learning model to update at least one parameter of the machine learning model, and (iii) by performing at least one generalization technique to improve a generalization capability of the machine learning model;

obtaining, by the computing system, itinerary data that identifies a planned shipment of an item of cargo from an origin location to a destination location by a shipping vehicle without identifying a transshipment location for the item of cargo;

generating, by a set of location devices respectively located on the shipping vehicle and a set of different shipping vehicles, vehicle location data describing a set of locations respectively of the shipping vehicle and the set of different shipping vehicles;

analyzing, by the machine learning model, the itinerary data and the vehicle location data to:

identify a plurality of probable transshipment locations that are sorted based on a respective frequency with which the respective probable transshipment location was previously used for transshipment,

determine that the vehicle location data indicates a first current or future presence of the shipping vehicle at a first probable transshipment location of the plurality of probable transshipment locations, the first probable transshipment location having the highest frequency use for transshipment of the plurality of probable transshipment locations, and

based on determining that the vehicle location data indicates the first current or future presence of the shipping vehicle at the first probable transshipment location, predict that the first probable transshipment location will be a transshipment location at which the item of cargo is transferred from the shipping vehicle to the different shipping vehicle;

inserting, by the computing system, the transshipment location into a data record associated with the planned shipment to produce an updated data record;

obtaining, by the computing system, vehicle schedule data that describes schedules of the shipping vehicle and the set of different shipping vehicles; and

refining, by the computing system using a schedule refinement machine learning model and based at least in part on the vehicle location data, the vehicle schedule data to produce refined vehicle schedule data.

2. The computer-implemented method of claim 1 , wherein the set of different shipping vehicles is related to the shipping vehicle via shared inclusion within a carrier network.

3. The computer-implemented method of claim 1 ,

the vehicle schedule data comprises:

shifting dates provided by the vehicle schedule data such that a scheduled location provided by the vehicle schedule data for the shipping vehicle aligns with a current location of the shipping vehicle.

4. The computer-implemented method of claim 1 , further comprising:

prior to analyzing the itinerary data and the vehicle location data:

obtaining, by the computing system, historical vehicle data that describes historical locations of the shipping vehicle; and

refining, by the computing system, the vehicle location data based at least in part on the historical vehicle data to produce refined vehicle location data.

5. The computer-implemented method of claim 1 , further comprising:

analyzing, by the machine learning model, the itinerary data and the vehicle location data to further predict one or more additional transshipment locations based at least in part on the transshipment location.

6. The computer-implemented method of claim 1 , further comprising:

predicting, by the computing system, a current location of the item of cargo based at least in part on the transshipment location.

7. The computer-implemented method of claim 1 , further comprising:

predicting, by the computing system, an updated discharge date of the item of cargo at the destination location based at least in part on the transshipment location.

8. The computer-implemented method of claim 7 , further comprising:

detecting, by the computing system, when the updated discharge date fails to satisfy a target discharge date;

and when the updated discharge date fails to satisfy a target discharge date:

providing, by the computing system, an alert to a shipper computing device associated with a shipper of the item of cargo.

9. The computer-implemented method of claim 1 , wherein the shipping vehicle comprises an origin shipping vehicle planned to load the item of cargo at the origin location.

10. A supply chain management computing system, comprising:

a set of location devices respectively located on a shipping vehicle and a set of different shipping vehicles, the set of location devices configured to generate vehicle location data describing a set of locations respectively of the shipping vehicle and the set of different shipping vehicles;

one or more processors; and

one or more non-transitory computer-readable media that collectively store instructions that, when executed by the one or more processors, cause the supply chain management computing system to perform operations comprising:

training a machine learning model (i) using a set of training data indicating a set of historical outcomes of a set previous shipments, (ii) based on a loss function evaluating a difference between a model prediction and a ground truth label being backpropagated through the machine learning model to update at least one parameter of the machine learning model, and (iii) by performing at least one generalization technique to improve a generalization capability of the machine learning model;

obtaining itinerary data that identifies a planned shipment of an item of cargo from an origin location to a destination location by a shipping vehicle without identifying a transshipment location for the item of cargo;

analyzing, by the machine learning model, the itinerary data and the vehicle location data to:

identify a plurality of probable transshipment locations that are sorted based on a respective frequency with which the respective probable transshipment location was previously used for transshipment,

determine that the vehicle location data indicates a first current or future presence of the shipping vehicle at a first probable transshipment location of the plurality of probable transshipment locations, the first probable transshipment location having the highest frequency use for transshipment of the plurality of probable transshipment locations, and

based on determining that the vehicle location data indicates the first current or future presence of the shipping vehicle at the first probable transshipment location, predict that the first probable transshipment location will be a transshipment location at which the item of cargo is transferred from the shipping vehicle to the different shipping vehicle;

inserting the transshipment location into a data record associated with the planned shipment to produce an updated data record;

obtaining vehicle schedule data that describes schedules of the shipping vehicle and the set of different shipping vehicles; and

refining, using a schedule refinement machine learning model and based at least in part on the vehicle location data, the vehicle schedule data to produce refined vehicle schedule data.

11. The supply chain management computing system of claim 10 , wherein the set of different shipping vehicles is related to the shipping vehicle via shared inclusion within a carrier network.

12. The supply chain management computing system of claim 10 , wherein the operations for refining the vehicle schedule data comprise operations for:

shifting dates provided by the vehicle schedule data such that a scheduled location provided by the vehicle schedule data for the shipping vehicle aligns with a current location of the shipping vehicle.

13. The supply chain management computing system of claim 10 , wherein the instructions that, when executed by the one or more processors, cause the supply chain management computing system to perform operations further comprising:

prior to analyzing the itinerary data and the vehicle location data:

obtaining historical vehicle data that describes historical locations of the shipping vehicle; and

refining the vehicle location data based at least in part on the historical vehicle data to produce refined vehicle location data.

14. One or more non-transitory computer-readable media that instructions that, when executed by one or more processors, cause a supply chain management computing system to perform operations comprising:

training a machine learning model (i) using a set of training data indicating a set of historical outcomes of a set previous shipments, (ii) based on a loss function evaluating a difference between a model prediction and a ground truth label being backpropagated through the machine learning model to update at least one parameter of the machine learning model, and (iii) by performing at least one generalization technique to improve a generalization capability of the machine learning model;

obtaining itinerary data that identifies a planned shipment of an item of cargo from an origin location to a destination location by a shipping vehicle without identifying a transshipment location for the item of cargo;

generating, by a set of location devices respectively located on the shipping vehicle and a set of different shipping vehicles, vehicle location data describing a set of locations respectively of the shipping vehicle and the set of different shipping vehicles;

analyzing, by the machine learning model, the itinerary data and the vehicle location data to:

identify a plurality of probable transshipment locations that are sorted based on a respective frequency with which the respective probable transshipment location was previously used for transshipment,

determine that the vehicle location data indicates a first current or future presence of the shipping vehicle at a first probable transshipment location of the plurality of probable transshipment locations, the first probable transshipment location having the highest frequency use for transshipment of the plurality of probable transshipment locations, and

based on determining that the vehicle location data indicates the first current or future presence of the shipping vehicle at the first probable transshipment location, predict that the first probable transshipment location will be a transshipment location at which the item of cargo is transferred from the shipping vehicle to a different shipping vehicle;

inserting the transshipment location into a data record associated with the planned shipment to produce an updated data record;

obtaining vehicle schedule data that describes schedules of the shipping vehicle and the set of different shipping vehicles; and

refining, using a schedule refinement machine learning model and based at least in part on the vehicle location data, the vehicle schedule data to produce refined vehicle schedule data.