US 12,412,101 B2
Predicting user state using machine learning
Michael O'Herlihy, San Francisco, CA (US); Rafiq Raziuddin Merchant, San Francisco, CA (US); Nirveek De, San Bruno, CA (US); and Jordan Allen Buettner, San Francisco, CA (US)
Assigned to Uber Technologies, Inc., San Francisco, CA (US)
Filed by Uber Technologies, Inc., San Francisco, CA (US)
Filed on Nov. 6, 2023, as Appl. No. 18/503,108.
Application 18/503,108 is a continuation of application No. 16/944,030, filed on Jul. 30, 2020, granted, now 11,847,573.
Application 16/944,030 is a continuation of application No. 15/367,139, filed on Dec. 1, 2016, granted, now 10,762,441, issued on Sep. 1, 2020.
Prior Publication US 2024/0144034 A1, May 2, 2024
This patent is subject to a terminal disclaimer.
Int. Cl. G06N 20/00 (2019.01); G06N 5/02 (2023.01)
CPC G06N 5/02 (2013.01) [G06N 20/00 (2019.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method for predicting user state in relation to a service, comprising:
receiving, from a first user device of a first user, a request to provide the service;
responsive to receiving the request to provide the service, receiving service data including service features and user features, the service features representing information about services provided to users and the user features representing information derived from current user activity data received from the first user device of the first user during provision of the service and past services requested by the first user;
generating a plurality of user feature values for the first user, wherein each user feature value is determined based on a difference between an average of values associated with the past services requested by the first user and a value derived from the current user activity data of the first user;
generating a prediction of a state of the first user by providing the generated plurality of user feature values for the first user and at least some of the service features as input to a machine learning model that has been trained using supervised learning to predict user state, wherein the prediction of the state of the first user is generated based on past services provided to the users, the past services provided to users including at least one past service provided to the first user and at least one past service provided to another user other than the first user;
dynamically altering one or more parameters associated with a manner in which the service is coordinated for the first user during provision of the service based on the prediction about the state of the first user; and
providing, via a wireless communication link, the altered one or more parameters to a provider client device of a provider of the service to cause a modification to the manner in which the service is coordinated for the first user.