| CPC G06Q 30/0631 (2013.01) [G06N 5/04 (2013.01); G06N 20/00 (2019.01)] | 20 Claims |
|
1. A method, comprising:
training, by a device, at least one of:
a first machine learning model, using historical customer data associated with previous customers of previous items, to group previous customers into groups of previous customers, or
a second machine learning model, using previous consumption occurrence measures and previous depth properties associated with the previous consumption occurrence measures, to determine previous associations between previous item data and the groups of previous customers;
receiving, by the device, customer data identifying one or more of actions, behaviors, or features associated with customers of items;
identifying, by the device, a set of customer characteristics based on the customer data;
grouping, by the device and based on processing the set of customer characteristics with the first machine learning model, the customers into groups of customers;
calculating, by the device, consumption occurrence measures for the items based on item data, wherein the consumption occurrence measures include a first probability of a first item being consumed when a second item has been consumed;
calculating, by the device, depth properties associated with the consumption occurrence measures, wherein the depth properties include a second probability of the first item being consumed when a plurality of items, including the second item, have been consumed in a time series;
determining, by the device and using the second machine learning model, associations, between the item data and the groups of customers, based on the consumption occurrence measures and the depth properties;
receiving, by the device and from a user device associated with a particular customer, particular customer data identifying one or more of actions, behaviors, or features, associated with the particular customer,
wherein the user device is different from the device;
identifying, by the device, particular customer characteristics for the particular customer based on the particular customer data;
assigning, by the device, the particular customer to a particular group, of the groups of customers, based on the particular customer characteristics;
generating, by the device, an item recommendation, recommending one of the items and for the particular customer, based on an association, between the particular group and the item data, of the determined associations;
providing, by the device, the item recommendation to the user device associated with the particular customer; and
retraining, by the device and based on the item recommendation, at least one of the first machine learning model or the second machine learning model,
wherein at least one of:
the first machine learning model is retrained to identify a pattern of the particular customer characteristics based on the item recommendation, or
the second machine learning model is retrained to determine a pattern of the determined associations, between the item data and the groups of customers, based on the item recommendation.
|