Predicting user state using machine learning
A system coordinates services between users and providers. The system trains a computer model to predict a user state of a user using data about past services. The prediction is based on data associated with a request submitted by a user. Request data can include current data about the user's behavior and information about the service that is independent of the particular user behavior or characteristics. The user behavior may be compared against the user's prior behavior to determine differences in the user behavior for this request and normal behavior of prior requests. The system can alter the parameters of a service based on the prediction about the state of the user requesting the service.
1. A computer-implemented method of data processing for predicting user state, comprising:
obtaining a set of training data comprising service data for each service of a plurality of services coordinated by a network system for a plurality of users including a first user, wherein the service data obtained for a service includes activity data corresponding to a user's interactions with a user device when requesting the service, and an indication of a user state of the user at the time of requesting the service;
training a computer model to predict the user state of the user using the obtained set of training data;
receiving current user activity data corresponding to a set of features of a request for service from a first user device of the first user and user profile data corresponding to a set of features of 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 the values of a feature from the user profile data for the first user and the value of the feature 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 a set of service features associated with the request for service from the first user as input to the trained computer model; and
altering one or more parameters associated with a manner in which the network system coordinates a service for the first user based on the prediction about the state of the first user.
2. The computer-implemented method of claim 1 , wherein the activity data comprises one or more of: data input accuracy, data input speed, interface interaction behavior, device angle, or walking speed.
3. The computer-implemented method of claim 1 , wherein the service data comprises one or more of: the first user's location when the first user requests the service, a time of day when the first user requests the service, or a day of the week when the first user requests the service.
4. The computer-implemented method of claim 1 , wherein altering one or more parameters associated with the manner in which the network system coordinates the service for the first user comprises alerting a provider about the state of the first user.
5. The computer-implemented method of claim 1 , wherein altering one or more parameters associated with the manner in which the network system coordinates the service for the first user comprises selecting an alternative meeting location at which a provider can meet with the first user.
6. The computer-implemented method of claim 1 , wherein altering one or more parameters associated with the manner in which the network system coordinates the service for the first user comprises presenting the first user with an alternative version of a map user interface.
7. The computer-implemented method of claim 1 , wherein altering one or more parameters associated with the manner in which the network system coordinates the service for the first user comprises assigning a provider for the service for the first user based on the provider's history of previous interactions with users that had the same state as the state of the first user.
8. The computer-implemented method of claim 1 , wherein, when the user profile of the first user includes less than a threshold amount of user data for the first user, the user profile of the first user includes user data of other users of the network system.
9. A non-transitory computer-readable storage medium storing computer program instructions executable by a processor of a system to perform steps for predicting user state comprising:
obtaining a set of training data comprising service data for each service of a plurality of services coordinated by a network system for a plurality of users including a first user, wherein the service data obtained for a service includes activity data corresponding to a user's interactions with a user device when requesting the service, and an indication of a user state of the user at the time of requesting the service;
training a computer model to predict the user state of the user using the obtained set of training data;
receiving current user activity data corresponding to a set of features of a request for service from a first user device of the first user and user profile data corresponding to a set of features of 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 the values of a feature from the user profile data for the first user and the value of the feature 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 a set of service features associated with the request for service from the first user as input to the trained computer model; and
altering one or more parameters associated with a manner in which the network system coordinates a service for the first user based on the prediction about the state of the first user.
10. The non-transitory computer-readable storage medium of claim 9 , wherein the activity data comprises one or more of: data input accuracy, data input speed, interface interaction behavior, device angle, or walking speed.
11. The non-transitory computer-readable storage medium of claim 9 , wherein the service data comprises one or more of: the first user's location when the first user requests the service, a time of day when the first user requests the service, or a day of the week when the first user requests the service.
12. The non-transitory computer-readable storage medium of claim 9 , wherein altering one or more parameters associated with the manner in which the network system coordinates the service for the first user comprises alerting a provider about the state of the first user.
13. The non-transitory computer-readable storage medium of claim 9 , wherein altering one or more parameters associated with the manner in which the network system coordinates the service for the first user comprises selecting an alternative meeting location at which a provider can meet with the first user.
14. The non-transitory computer-readable storage medium of claim 9 , wherein altering one or more parameters associated with the manner in which the network system coordinates the service for the first user comprises presenting the first user with an alternative version of a map user interface.
15. The non-transitory computer-readable storage medium of claim 9 , wherein altering one or more parameters associated with the manner in which the network system coordinates the service for the first user comprises assigning a provider for the service for the first user based on the provider's history of previous interactions with users that had the same state as the state of the first user.
16. The non-transitory computer-readable storage medium of claim 9 , wherein, when the user profile of the first user includes less than a threshold amount of user data for the first user, the user profile of the first user includes user data of other users of the network system.