User interface that pre-populates items in an order module for a user of an online concierge system using a prediction model

Abstract

An online concierge system maintains historical orders received from a user that include one or more items. For items included in one more historical orders, the online concierge system determines an interval between orders including an item, providing an indication of a frequency with which the user orders the item. When the online concierge system receives a request to create an order from the user, in response to an amount of time between a most recently received order including the item and a time when the request was received is within a threshold duration of the interval between orders including the item, the online concierge system selects an item from a category including the item. The selected item may be the item or an alternative item in the category. Subsequently, the online concierge system displays an interface for the user that is prepopulated with information identifying the selected item.

What is claimed is: 
     
         1 . A method comprising, at an online concierge system comprising at least one processor and memory:
 receiving, by the online concierge system, a request to create an order associated with a user;   retrieving, by the online concierge system, historical orders associated with the user, each historical order including one or more items selected by the user and associated with a time when a historical order was received;   for each of at least a set of items included in at least one historical order:
 determining, by the online concierge system, time intervals between pairs of consecutive historical orders including an item of the set; 
 determining, by the online concierge system, an interval between orders including the item of the set from the determined time intervals; and 
 storing, at the online concierge system, the interval between orders in association with the user and in association with the item of the set; 
   identifying, by the online concierge system, an item having an interval between orders greater than an amount of time between a time of a most recent previously received order including the item and a time when the request to create the order was received that is within a threshold duration of the interval between orders stored in association with the user and with the item;   determining, by the online concierge system, a category of items including the identified item;   applying, by the online concierge system, a purchase model for predicting a probability of the user purchasing an item in the determined category, comprising:
 accessing, by the online concierge system, the purchase model, wherein the purchase model comprises a machine learning model trained by:
 accessing a training dataset that describes attributes of an item of a determined category and historical purchase information of the item included in the historical orders associated with the user; 
 performing a learning process on the purchase model to train weights of the model based on the training dataset; 
 updating the training dataset with recent purchase information; and 
 retraining the purchase model based on the updated training dataset; 
 
   selecting, by the online concierge system, an item from the determined category based on the predicted probability of the user purchasing the item in the determined category; and   generating, by the online concierge system, an interface for the order for which the request was received that includes information identifying the selected item from the determined category.   
     
     
         2 . The method of  claim 1 , wherein identifying, by the online concierge system, an item having an interval between orders greater than an amount of time between a time of a most recent previously received order including the item and a time when the request to create the order was received that is within a threshold duration of the interval between orders stored in association with the user and with the item comprises:
 identifying an item for which the amount of time between the time of the most recent previously received order including the item and the time when the request to create the order was received is greater than the interval between orders stored in association with the user and with the item.   
     
     
         3 . The method of  claim 1 , wherein selecting, by the online concierge system, an item from the determined category comprises:
 selecting an item having a maximum probability of being purchased by the user from the determined category.   
     
     
         4 . The method of  claim 1 , wherein selecting, by the online concierge system, an item from the determined category comprises:
 determining differences between a probability of the user purchasing the identified item and probabilities of the user purchasing each alternative item in the determined category;   selecting an alternative item in the determined category having a minimum difference between the probability of the user purchasing the identified item and the probability of the user purchasing the alternative item in the determined category.   
     
     
         5 . The method of  claim 1 , wherein selecting, by the online concierge system, an item from the determined category comprises:
 determining a predicted availability of the identified item at a warehouse identified by the request;   in response to the predicted availability of the identified item at the warehouse being less than a threshold availability, determining a probability of the user purchasing each alternative item in the determined category; and   selecting an alternative item in the determined category having a maximum probability of being purchased by the user.   
     
     
         6 . The method of  claim 1 , wherein selecting, by the online concierge system, an item from the determined category comprises:
 determining a predicted availability of the identified item at a warehouse identified by the request;   in response to the predicted availability of the identified item at the warehouse being less than a threshold availability, retrieving stored information associated with the user by the online concierge system identifying one or more replacement items for the identified item; and   selecting a replacement item.   
     
     
         7 . The method of  claim 1 , wherein selecting, by the online concierge system, an item from the determined category comprises:
 determining a predicted availability of the identified item at a warehouse identified by the request;   in response to the predicted availability of the identified item at the warehouse being less than a threshold availability, retrieving stored information associated with the user by the online concierge system identifying one or more replacement items for the identified item;   determining a probability of the user purchasing each of the replacement items; and   selecting a replacement item having a maximum probability of being purchased by the user.   
     
     
         8 . A computer program product comprising a non-transitory computer readable storage medium having instructions encoded thereon that, when executed by a processor, cause the processor to:
 receive a request to create an order associated with a user;   retrieve historical orders associated with the user, each historical order including one or more items selected by the user and associated with a time when a historical order was received;   for each of at least a set of items included in at least one historical order:
 determine time intervals between pairs of consecutive historical orders including an item of the set; 
 determine an interval between orders including the item of the set from the determined time intervals; and 
 store the interval between orders in association with the user and in association with the item of the set; 
   identify an item having an interval between orders greater than an amount of time between a time of a most recent previously received order including the item and a time when the request to create the order was received that is within a threshold duration of the interval between orders stored in association with the user and with the item;   determine a category of items including the identified item;   apply a purchase model for predicting a probability of the user purchasing an item in the determined category, comprising:
 accessing the purchase model, wherein the purchase model comprises a machine learning model trained by:
 accessing a training dataset that describes attributes of an item of a determined category and historical purchase information of the item included in the historical orders associated with the user; 
 performing a learning process on the purchase model to train weights of the model based on the training dataset; 
 updating the training dataset with recent purchase information; and 
 retraining the purchase model based on the updated training dataset; 
 
   select an item from the determined category based on the predicted probability of the user purchasing the item in the determined category; and   generate an interface for the order for which the request was received that includes information identifying the selected item from the determined category.   
     
     
         9 . The computer program product of  claim 8 , wherein the instruction to identify an item having an interval between orders greater than an amount of time between a time of a most recent previously received order including the item and a time when the request to create the order was received that is within a threshold duration of the interval between orders stored in association with the user and with the item comprises:
 identifying an item for which the amount of time between the time of the most recent previously received order including the item and the time when the request to create the order was received is greater than the interval between orders stored in association with the user and with the item.   
     
     
         10 . The computer program product of  claim 8 , wherein the instruction to select an item from the determined category comprises:
 selecting an item having a maximum probability of being purchased by the user from the determined category.   
     
     
         11 . The computer program product of  claim 8 , wherein the instruction to select an item from the determined category comprises:
 determining differences between a probability of the user purchasing the identified item and probabilities of the user purchasing each alternative item in the determined category;   selecting an alternative item in the determined category having a minimum difference between the probability of the user purchasing the identified item and the probability of the user purchasing the alternative item in the determined category.   
     
     
         12 . The computer program product of  claim 8 , wherein the instruction to select an item from the determined category comprises:
 determining a predicted availability of the identified item at a warehouse identified by the request;   in response to the predicted availability of the identified item at the warehouse being less than a threshold availability, determining a probability of the user purchasing each alternative item in the determined category; and   selecting an alternative item in the determined category having a maximum probability of being purchased by the user.   
     
     
         13 . The computer program product of  claim 8 , wherein the instruction to select an item from the determined category comprises:
 determining a predicted availability of the identified item at a warehouse identified by the request;   in response to the predicted availability of the identified item at the warehouse being less than a threshold availability, retrieving stored information associated with the user identifying one or more replacement items for the identified item; and   selecting a replacement item.   
     
     
         14 . The computer program product of  claim 8 , wherein the instruction to select an item from the determined category comprises:
 determining a predicted availability of the identified item at a warehouse identified by the request;   in response to the predicted availability of the identified item at the warehouse being less than a threshold availability, retrieving stored information associated with the user identifying one or more replacement items for the identified item;   determining a probability of the user purchasing each of the replacement items; and   selecting a replacement item having a maximum probability of being purchased by the user.   
     
     
         15 . A computer system comprising:
 a processor; and   a non-transitory computer-readable storage medium having instructions that, when executed by the processor, cause the computer system to perform steps comprising:
 receiving a request to create an order associated with a user; 
 retrieving historical orders associated with the user, each historical order including one or more items selected by the user and associated with a time when a historical order was received; 
 for each of at least a set of items included in at least one historical order:
 determining time intervals between pairs of consecutive historical orders including an item of the set; 
 determining an interval between orders including the item of the set from the determined time intervals; and 
 storing the interval between orders in association with the user and in association with the item of the set; 
 
 identifying an item having an interval between orders greater than an amount of time between a time of a most recent previously received order including the item and a time when the request to create the order was received that is within a threshold duration of the interval between orders stored in association with the user and with the item; 
 determining a category of items including the identified item; 
 applying a purchase model for predicting a probability of the user purchasing an item in the determined category, comprising:
 accessing the purchase model, wherein the purchase model comprises a machine learning model trained by:
 accessing a training dataset that describes attributes of an item of a determined category and historical purchase information of the item included in the historical orders associated with the user; 
 performing a learning process on the purchase model to train weights of the model based on the training dataset; 
 updating the training dataset with recent purchase information; and 
 retraining the purchase model based on the updated training dataset; 
 
 selecting an item from the determined category based on the predicted probability of the user purchasing the item in the determined category; and 
 generating an interface for the order for which the request was received that includes information identifying the selected item from the determined category. 
 
   
     
     
         16 . The computer system of  claim 15 , wherein selecting an item from the determined category comprises:
 selecting an item having a maximum probability of being purchased by the user from the determined category.   
     
     
         17 . The computer system of  claim 15 , wherein selecting an item from the determined category comprises:
 determining differences between a probability of the user purchasing the identified item and probabilities of the user purchasing each alternative item in the determined category;   selecting an alternative item in the determined category having a minimum difference between the probability of the user purchasing the identified item and the probability of the user purchasing the alternative item in the determined category.   
     
     
         18 . The computer system of  claim 15 , wherein selecting an item from the determined category comprises:
 determining a predicted availability of the identified item at a warehouse identified by the request;   in response to the predicted availability of the identified item at the warehouse being less than a threshold availability, determining a probability of the user purchasing each alternative item in the determined category; and   selecting an alternative item in the determined category having a maximum probability of being purchased by the user.   
     
     
         19 . The computer system of  claim 15 , wherein selecting an item from the determined category comprises:
 determining a predicted availability of the identified item at a warehouse identified by the request;   in response to the predicted availability of the identified item at the warehouse being less than a threshold availability, retrieving stored information associated with the user identifying one or more replacement items for the identified item; and   selecting a replacement item.   
     
     
         20 . The computer system of  claim 15 , wherein selecting an item from the determined category comprises:
 determining a predicted availability of the identified item at a warehouse identified by the request;   in response to the predicted availability of the identified item at the warehouse being less than a threshold availability, retrieving stored information associated with the user identifying one or more replacement items for the identified item;   determining a probability of the user purchasing each of the replacement items; and   selecting a replacement item having a maximum probability of being purchased by the user.