Systems and methods for automatic product usage model training and prediction

Abstract

Disclosed are methods, systems, and non-transitory computer-readable medium for executing neural network training for dynamically predicting apparel wearability. For example, a method may include generating a training data set comprising one or more historical data attributes of previously shipped apparel, training a neural network based on the training data set to configure one or more trained models to output a metric for any pair of a unique user identifier and a unique apparel identifier, storing one or more trained model objects, collecting prediction data comprising at least one prediction pair including a unique user identifier and a unique apparel identifier, predicting one or more predictive wearability metrics indicative of propensity to wear, dynamically generating one or more match pairs, and determining a match wearability metric for each of the one or more match pairs based on the predicted one or more predictive wearability metrics.

1 - 20 . (canceled) 
     
     
         21 . A computer-implemented method for executing machine learning training for dynamically predicting product usage in an electronics transactions platform, the method comprising:
 generating, by one or more processors, a training data set comprising one or more historical data attributes of previously used products, each of the historical data attributes being linked to a pair of a unique user identifier and a unique product identifier used in the electronic transactions platform to form feature vectors for each pair of a unique user identifier and a unique product identifier;   training, by the one or more processors, a machine learning model based on the training data set to configure one or more trained models to output a metric for any pair of a unique user identifier and a unique product identifier used in the electronic transactions platform;   storing, by the one or more processors, the one or more trained models as one or more trained model objects at a database of the electronic transactions platform;   collecting, by the one or more processors, prediction data comprising at least one prediction pair including a unique user identifier and a unique product identifier used in the electronic transactions platform, each prediction pair corresponding to product saved for a future transaction, purchased, or returned through the electronic transactions platform;   executing, by the one or more processors, the stored one or more trained model objects with the prediction data to determine one or more predictive usage metrics indicative of a propensity of a user to use product saved for a future transaction, purchased, or returned;   periodically storing the one or more predictive usage metrics at a metrics database of the electronic transactions platform, wherein storing the one or more predictive usage metrics includes storing a first predictive usage metric associated with a first prediction pair, the first prediction pair including a first unique user identifier and a first unique product identifier, the first predictive usage metric indicating a propensity of a user associated with the first unique user identifier to use product associated with the first unique product identifier;   dynamically generating, by the one or more processors, one or more match pairs, each match pair including a unique user identifier and a unique product identifier used in the electronic transactions platform; and   periodically determining, by the one or more processors, a match usage metric for each of the one or more match pairs based on the stored one or more predictive usage metrics, wherein determining the match usage metric for each of the one or more match pairs includes: determining a first match pair including the first unique user identifier and the first unique product identifier, retrieving the stored first predictive usage metric associated with the first unique user identifier and the first unique product identifier, and storing the retrieved first predictive usage metric as a first match usage metric for the first match pair, the first match usage metric indicating the propensity of the user associated with the first unique user identifier to use product associated with the first unique product identifier.   
     
     
         22 . The method of  claim 21 , wherein the generating the training data set further comprises determining one or more distributions of fit ratings corresponding to a unique product identifier. 
     
     
         23 . The method of  claim 21 , wherein the storing the one or more predictive usage metrics includes limiting the storing the one or more predictive usage metrics based on preset conditions. 
     
     
         24 . The method of  claim 21 , wherein the metrics database is a database configured to index, retrieve, and search a plurality of data files comprising the one or more predictive usage metrics. 
     
     
         25 . The method of  claim 23 , wherein the storing the one or more predictive usage metrics includes cleaning the stored one or more predictive usage metrics based on additional preset conditions. 
     
     
         26 . The method of  claim 21 , wherein the determining the match usage metric for each of the one or more match pairs further comprises retrieving a predictive usage metric among the one or more predictive usage metrics stored at the metrics database. 
     
     
         27 . The method of  claim 21 , wherein the dynamically generating the one or more match pairs is performed in response to receiving one or more replenishment identifiers. 
     
     
         28 . A computer system for executing machine learning training for dynamically predicting product usage in an electronics transactions platform, the computer system comprising:
 a memory having processor-readable instructions stored therein; and   at least one processor configured to access the memory and execute the processor-readable instructions, which when executed by the at least one processor configures the at least one processor to perform a plurality of functions, including functions for:
 generating a training data set comprising one or more historical data attributes of previously used products, each of the historical data attributes being linked to a pair of a unique user identifier and a unique product identifier used in the electronics transactions platform to form feature vectors for each pair of a unique user identifier and a unique product identifier; 
 training a machine learning model based on the training data set to configure one or more trained models to output a metric for any pair of a unique user identifier and a unique product identifier used in the electronics transactions platform; 
 storing the one or more trained models as one or more trained model objects at a database of the electronics transactions platform; 
 collecting prediction data comprising at least one prediction pair including a unique user identifier and a unique product identifier used in the electronic transactions platform, each prediction pair corresponding to product saved for a future transaction, purchased, or returned through the electronic transactions platform; 
 executing the stored one or more trained model objects with the prediction data to determine one or more predictive usage metrics indicative of a propensity of a user to use product saved for a future transaction, purchased, or returned; 
 periodically storing the one or more predictive usage metrics at a metrics database of the electronic transactions platform, wherein storing the one or more predictive usage metrics includes storing a first predictive usage metric associated with a first prediction pair, the first prediction pair including a first unique user identifier and a first unique product identifier, the first predictive usage metric indicating a propensity of a user associated with the first unique user identifier to use product associated with the first unique product identifier; 
 dynamically generating one or more match pairs, each match pair including a unique user identifier and a unique product identifier used in the electronic transactions platform; and 
 periodically determining a match usage metric for each of the one or more match pairs based on the stored one or more usage metrics, wherein determining the match usage metric for each of the one or more match pairs includes: determining a first match pair including the first unique user identifier and the first unique product identifier, retrieving the stored first predictive usage metric associated with the first unique user identifier and the first unique product identifier, and storing the retrieved first predictive usage metric as a first match usage metric for the first match pair, the first match usage metric indicating the propensity of the user associated with the first unique user identifier to use product associated with the first unique product identifier. 
   
     
     
         29 . The system of  claim 28 , wherein the generating the training data set further comprises determining one or more distributions of fit ratings corresponding to a unique product identifier. 
     
     
         30 . The system of  claim 28 , wherein the storing the one or more predictive usage metrics includes limiting the storing the one or more predictive usage metrics based on preset conditions. 
     
     
         31 . The system of  claim 28 , wherein the metrics database is a database configured to index, retrieve, and search a plurality of data files comprising the one or more predictive usage metrics. 
     
     
         32 . The system of  claim 30 , wherein the storing the one or more predictive usage metrics includes cleaning the stored one or more predictive usage metrics based on additional preset conditions. 
     
     
         33 . The system of  claim 28 , wherein the determining the match usage metric for each of the one or more match pairs further comprises retrieving a predictive usage metric among the one or more predictive usage metrics stored at the metrics database. 
     
     
         34 . The system of  claim 28 , wherein the dynamically generating the one or more match pairs is performed in response to receiving one or more replenishment identifiers. 
     
     
         35 . A non-transitory computer-readable medium containing instructions for executing machine learning training for dynamically predicting product usage in an electronics transactions platform, comprising:
 generating a training data set comprising one or more historical data attributes of previously used products, each of the historical data attributes being linked to a pair of a unique user identifier and a unique product identifier used in the electronic transactions platform to form feature vectors for each pair of a unique user identifier and a unique product identifier;   training a machine learning model based on the training data set to configure one or more trained models to output a metric for any pair of a unique user identifier and a unique product identifier used in the electronic transactions platform;   storing the one or more trained models as one or more trained model objects at a database of the electronic transactions platform;   collecting prediction data comprising at least one prediction pair including a unique user identifier and a unique product identifier used in the electronic transactions platform, each prediction pair corresponding to product saved for a future transaction, purchased, or returned through the electronic transactions platform;   executing the stored one or more trained model objects with the prediction data to determine one or more predictive usage metrics indicative of a propensity of a user to use product saved for a future transaction, purchased, or returned;   periodically storing the one or more predictive usage metrics at a metrics database of the electronic transactions platform, wherein storing the one or more predictive usage metrics includes storing a first predictive usage metric associated with a first prediction pair, the first prediction pair including a first unique user identifier and a first unique product identifier, the first predictive usage metric indicating a propensity of a user associated with the first unique user identifier to use product associated with the first unique product identifier;   dynamically generating one or more match pairs, each match pair including a unique user identifier and a unique product identifier used in the electronic transactions platform; and   periodically determining a match usage metric for each of the one or more match pairs based on the stored one or more usage metrics, wherein determining the match usage metric for each of the one or more match pairs includes: determining a first match pair including the first unique user identifier and the first unique product identifier, retrieving the stored first predictive usage metric associated with the first unique user identifier and the first unique product identifier, and storing the retrieved first predictive usage metric as a first match usage metric for the first match pair, the first match usage metric indicating the propensity of the user associated with the first unique user identifier to use product associated with the first unique product identifier.   
     
     
         36 . The non-transitory computer-readable medium of  claim 35 , wherein the generating the training data set further comprises determining one or more distributions of fit ratings corresponding to a unique product identifier. 
     
     
         37 . The non-transitory computer-readable medium of  claim 35 , wherein determining the match usage metric for each of the one or more match pairs based on the predicted one or more predictive usage metrics comprises:
 retrieving the stored one or more predictive usage metrics from the metrics database.   
     
     
         38 . The non-transitory computer-readable medium of  claim 35 , wherein the metrics database is a database configured to index, retrieve, and search a plurality of data files comprising the one or more predictive usage metrics. 
     
     
         39 . The non-transitory computer-readable medium of  claim 35 , wherein the storing the one or more predictive usage metrics includes limiting the storing the one or more predictive usage metrics based on preset conditions. 
     
     
         40 . The non-transitory computer-readable medium of  claim 35 , wherein the dynamically generating the one or more match pairs is performed in response to receiving one or more replenishment identifiers.