Systems and methods for AI-based detection of delays in a shipping network
Embodiments provide systems, methods and computer program products for artificial intelligence-based detection of delays in a shipping network that include training a machine learning model using a training set of in-flight snapshots to infer, based on a set of predictive features, whether a shipment having an estimated delivery dates will meet the estimated delivery date. The machine learning model represents a set of shipment statuses, a set of timings relative estimated delivery dates, and a set of shipment outcomes of the plurality of historical shipments.
1. A computer program product comprising a non-transitory, computer-readable medium storing thereon a set of computer-executable instructions, the set of computer-executable instructions comprising instructions for:
collecting a set of records representing a plurality of historical shipments, each historical shipment in the plurality of historical shipments having a historical shipment estimated delivery date and a historical shipment actual delivery date;
for each historical shipment in the plurality of historical shipments, accessing a training set of in-flight snapshots from multiple different snapshot times for that historical shipment, wherein each in-flight snapshot in the training set of in-flight snapshots represents a respective snapshot time while that historical shipment was in-flight and before that historical shipments was delivered, and wherein each in-flight snapshot in the training set of in-flight snapshots has an associated time remaining to estimated delivery date based on the respective snapshot time and the historical shipment estimated delivery date of that historical shipment;
for each training set of in-flight snapshots respectively corresponding to the plurality of historical shipments:
extracting a set of input features respectively corresponding to that training set of in-flight snapshots, and
generating a feature vector based on the set of input features;
training a machine learning model using the feature vectors that were respectively generated for the training sets of in-flight snapshots;
tracking shipment information for an in-progress shipment, the shipment information comprising status information and an associated estimated delivery date for the in-progress shipment, wherein the shipment information is tracked during shipment of the in-progress shipment; and
executing the machine learning model on the shipment information for the in-progress shipment to output an inference for the in-progress shipment, the inference indicating a probability of whether delivery of the in-progress shipment will be delayed.
2. The computer program product of claim 1 , wherein the set of computer-executable instructions further comprises instructions for:
based on a determination that the probability of whether delivery of the in-progress shipment will be delayed exceeds a threshold, marking the in-progress shipment as being at risk of missing the associated estimated delivery date for the in-progress shipment.
3. The computer program product of claim 2 , further comprising storing the shipment information and the probability of whether delivery of the in-progress shipment will be delayed for the in-progress shipment in a database indexed for a web application.
4. The computer program product of claim 1 , wherein the feature vector for each training set of in-flight snapshots comprises a current status feature and a time to estimated delivery date feature.
5. The computer program product of claim 4 , wherein the feature vector for each training set of in-flight snapshots includes a carrier feature.
6. The computer program product of claim 5 , wherein the feature vector for each training set of in-flight snapshots includes a class of service feature.
7. The computer program product of claim 6 , wherein the feature vector for each training set of in-flight snapshots includes a package weight feature.
8. A computer-implemented method, implemented by one or more processors, comprising:
collecting a set of records representing a plurality of historical shipments, each historical shipment in the plurality of historical shipments having a historical shipment estimated delivery date and a historical shipment actual delivery date;
for each historical shipment in the plurality of historical shipments, accessing a training set of in-flight snapshots from multiple different snapshot times for that historical shipment, wherein each in-flight snapshot in the training set of in-flight snapshots represents a respective snapshot time while that historical shipment was in-flight and before that historical shipments was delivered, and wherein each in-flight snapshot in the training set of in-flight snapshots has an associated time remaining to estimated delivery date based on the respective snapshot time and the historical shipment estimated delivery date of that historical shipment;
for each training set of in-flight snapshots respectively corresponding to the plurality of historical shipments:
extracting a set of input features respectively corresponding to that training set of in-flight snapshots, and
generating a feature vector based on the set of input features;
training a machine learning model using the feature vectors that were respectively generated for the training sets of in-flight snapshots;
tracking shipment information for an in-progress shipment, the shipment information comprising status information and an associated estimated delivery date for the in-progress shipment, wherein the shipment information is tracked during shipment of the in-progress shipment; and
executing the machine learning model on the shipment information for the in-progress shipment to output an inference for the in-progress shipment, the inference indicating a probability of whether delivery of the in-progress shipment will be delayed.
9. The computer-implemented method of claim 8 , further comprising:
based on a determination that the probability of whether delivery of the in-progress shipment will be delayed for the in-progress shipment exceeds a threshold, marking the in-progress shipment as being at risk of missing the associated estimated delivery date for the in-progress shipment.
10. The computer-implemented method of claim 9 , further comprising storing the shipment information and the delay risk score for the in-progress shipment in a database indexed for a web application.
11. The computer-implemented method of claim 9 , wherein the feature vector for each training set of in-flight snapshots comprises a current status feature and a time to estimated delivery date feature.
12. The computer-implemented method of claim 11 , wherein the feature vector for each training set of in-flight snapshots includes a carrier feature.
13. The computer-implemented method of claim 12 , wherein the feature vector for each training set of in-flight snapshots includes a class of service feature.
14. The computer-implemented method of claim 13 , wherein the feature vector for each training set of in-flight snapshots includes a package weight feature.