Using a trained model to generate action recommendations by predicting metrics related to items ordered at an online system

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 perform steps comprising:

• operating one or more weight sensors, one or more cameras and a computer vision of a smart shopping cart that is integrated into a computer system and is in communication with the computer system to gather replacement data with information about an item and a different item,
• wherein the replacement data is gathered via the one or more weight sensors, the one or more cameras and the computer vision when a user of the computer system uses the smart shopping cart at a location of a retailer during a current session of the user and replaces the item that was initially placed in the smart shopping cart with the different item that is currently placed in the smart shopping cart, and
• wherein the replacement data gathered via the one or more weight sensors, the one or more cameras and the computer vision includes information that the item and the different item are of a same type, information that the different item is placed in the smart shopping cart at a time after a removal of the item from the smart shopping cart, and information that content of the smart shopping cart was not changed after the removal of the item and before a placement of the different item in the smart shopping cart;
• uploading, from the smart shopping cart and via a network, the replacement data;
• responsive to uploading the replacement data, accessing a metric prediction model, wherein the metric prediction model is a machine-learning model trained to predict at least one of an irreplaceability feature for the item or a brand stickiness feature for the item;
• applying the metric prediction model to the replacement data to generate at least one of an irreplaceability score for the item indicative of the irreplaceability feature or a brand stickiness score for the item indicative of the brand stickiness feature;
• accessing an item replacement model, wherein the item replacement model is a machine-learning model trained to identify a list of one or more candidate replacement items for replacing the item;
• applying the item replacement model to at least one of the irreplaceability score or the brand stickiness score to generate the list of one or more candidate replacement items;
• and
• causing the smart shopping cart to display a user interface with information about a replacement item selected from the list prompting the user to place the replacement item in the smart shopping cart during the current session of the user.

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:
• operating one or more weight sensors, one or more cameras and a computer vision of a smart shopping cart that is integrated into the computer system and is in communication with the computer system to gather replacement data with information about an item and a different item,
• wherein the replacement data is gathered via the one or more weight sensors, the one or more cameras and the computer vision when a user of the computer system uses the smart shopping cart at a location of a retailer during a current session of the user and replaces the item that was initially placed in the smart shopping cart with the different item that is currently placed in the smart shopping cart, and
• wherein the replacement data gathered via the one or more weight sensors, the one or more cameras and the computer vision includes information that the item and the different item are of a same type, information that the different item is placed in the smart shopping cart at a time after a removal of the item from the smart shopping cart, and information that content of the smart shopping cart was not changed after the removal of the item and before a placement of the different item in the smart shopping cart;
• uploading, from the smart shopping cart and via a network, the replacement data;
• responsive to uploading the replacement data, accessing a metric prediction model, wherein the metric prediction model is a machine-learning model trained to predict at least one of an irreplaceability feature for the item or a brand stickiness feature for the item;
• applying the metric prediction model to the replacement data to generate at least one of an irreplaceability score for the item indicative of the irreplaceability feature or a brand stickiness score for the item indicative of the brand stickiness feature;
• accessing an item replacement model, wherein the item replacement model is a machine-learning model trained to identify a list of one or more candidate replacement items for replacing the item;
• applying the item replacement model to at least one of the irreplaceability score or the brand stickiness score to generate the list of one or more candidate replacement items;
• and
• causing the smart shopping cart to display a user interface with information about a replacement item selected from the list prompting the user to place the replacement item in the smart shopping cart during the current session of the user.