Systems and methods for prioritizing orders

1 . A method comprising:

building a machine learning (ML model) by:

generating, by an order prioritization service, a modeling dataset from a corpus of historical order fulfillment data from a plurality of orders from a plurality of customers using a natural language process;

implementing through a deep neural network, by an order prioritization service, one of a pairwise ranking process and a pointwise ranking process, wherein, when implementing the pairwise ranking process, the deep neural network takes as input features of two different orders and outputs a respective priority score for each of the two orders;

training, by the order prioritization service, the deep neural network to predict priority scores for orders to be fulfilled using a plurality of training samples, each training sample generated from modeling dataset, each training sample of the plurality of training samples to adjust weights in the deep neural network, wherein training the deep neural network includes inputting different portions of the modeling dataset and comparing predictions of customer actions with target values of the training samples to adjust weights in the deep neural network;

receiving, by the order prioritization service, information regarding a plurality of orders that need to be fulfilled;

determining, by the order prioritization service, for each one of the plurality of orders, one or more relevant features from the information regarding the order, the one or more relevant features influencing prediction of an order priority;

generating, by the order prioritization service, a feature vector that represents the one or more relevant features;

inputting, by the order prioritization service, the feature vector to the ML model;

predicting, by the order prioritization service using the ML model, a priority score for each one of the plurality of orders based on the determined one or more relevant features;

ranking, by the order prioritization service, the plurality of orders based on their respective priority scores; and

queuing, by the order prioritization service, the plurality of orders based on the ranking.

2 . The method of claim 1 , wherein the deep neural network includes a multilayer perceptron (MLP).

3 . The method of claim 1 , wherein the deep neural network is trained using supervised learning with the historical order fulfillment data including an actual priority of each order.

4 . The method of claim 1 , wherein the pairwise ranking model includes a gradient boosting forest.

5 . The method of claim 1 , wherein the pointwise ranking model includes one of a linear regression model or a multiple regression model.

6 . A system comprising:

one or more non-transitory machine-readable mediums configured to store instructions; and

one or more processors configured to execute the instructions stored on the one or more non-transitory machine-readable mediums, wherein execution of the instructions causes the one or more processors to carry out a process comprising:

building a machine learning (ML model) by:

generating a modeling dataset from a corpus of historical order fulfillment data from a plurality of orders from a plurality of customers using a natural language process;

implementing through a deep neural network one of a pairwise ranking process and a pointwise ranking process wherein, when implementing the pairwise ranking process, the deep neural network takes as input features of two different orders and output a respective priority score for each of the two orders;

training the deep neural network to predict priority scores for orders to be fulfilled using a plurality of training samples, each training sample generated from modeling dataset, each training sample of the plurality of training samples to adjust weights in the deep neural network, wherein training the deep neural network includes inputting different portions of the modeling dataset and comparing predictions of customer actions with target values of the training samples to adjust weights in the deep neural network;

receiving information regarding a plurality of orders that need to be fulfilled;

determining, for each one of the plurality of orders, one or more relevant features from the information regarding the order, the one or more relevant features influencing prediction of an order priority;

generating a feature vector that represents the one or more relevant features;

inputting the feature vector to the ML model;

predicting, using the ML model, a priority score for each one of the plurality of orders based on the determined one or more relevant features;

ranking the plurality of orders based on their respective priority scores;

queuing the plurality of orders based on the ranking.

7 . The system of claim 6 , wherein the deep neural network includes a multilayer perceptron (MLP).

8 . The system of claim 6 , wherein the deep neural network is trained using supervised learning with the historical order fulfillment data including an actual priority of each order.

9 . The system of claim 6 , wherein the pairwise ranking model includes a gradient boosting forest.

10 . The system of claim 6 , wherein the pointwise ranking model includes one of a linear regression model or a multiple regression model.

11 . A non-transitory machine-readable medium encoding instructions that when executed by one or more processors cause a process to be carried out, the process including:

building a machine learning (ML model) by:

generating a modeling dataset from a corpus of historical order fulfillment data from a plurality of orders from a plurality of customers using a natural language process;

implementing through a deep neural network at least one of a pairwise ranking process and a pointwise ranking process wherein, when implementing the pairwise ranking process, the deep neural network takes as input features of two different orders and outputs a respective priority score for each of the two orders;

training the deep neural network to predict priority scores for orders to be fulfilled using a plurality of training samples, each training sample generated from modeling dataset, each training sample of the plurality of training samples to adjust weights in the deep neural network, wherein training the deep neural network includes inputting different portions of the modeling dataset and comparing predictions of customer actions with target values of the training samples to adjust weights in the deep neural network;

receiving information regarding a plurality of orders that need to be fulfilled;

determining, for each one of the plurality of orders, one or more relevant features from the information regarding the order, the one or more relevant features influencing prediction of an order priority;

generating a feature vector that represents the one or more relevant features;

inputting the feature vector to the ML model;

predicting, using the ML model, a priority score for each one of the plurality of orders based on the determined one or more relevant features;

ranking the plurality of orders based on their respective priority scores queuing the plurality of orders based on the ranking.

12 . The machine-readable medium of claim 11 , wherein the deep neural network is trained using supervised learning with the historical order fulfillment data including an actual priority of each order.

13 . The machine readable medium of claim 11 wherein the process further comprises:

implementing a pairwise ranking process and a pointwise ranking process;

generating a normalized discounted cumulative gain measure for each of the pairwise ranking process and the pointwise ranking process;

selecting one of the pairwise ranking process and the pointwise ranking process based on the normalized discounted cumulative gain measure; and

implementing the selected one of the pairwise ranking process and the pointwise ranking process through the deep neural network.

14 . The machine readable medium of claim 13 wherein selecting one of the pairwise ranking process and the pointwise ranking process is based on a highest normalized discounted cumulative gain measure.

15 . The machine readable medium of claim 13 wherein selecting one of the pairwise ranking process and the pointwise ranking process is based on the normalized discounted cumulative gain measure satisfying a predetermined threshold.