Machine learning-based resource customization to increase user satisfaction

Patent metadata

Field	Value
Publication number	US-11308428-B2
Application number	US-201916505992-A
Country	US
Kind code	B2
Filing date	Jul 9, 2019
Priority date	Jul 9, 2019
Publication date	Apr 19, 2022
Grant date	Apr 19, 2022

Abstract

Customizing computing resource allocation based on machine learning is provided. A plurality of current users of a service are evaluated based on attributes of each current user. A user value score is generated for each current user based on the attributes. The plurality of current users is clustered into a plurality of user groups based on the user value score. A likelihood of each prospective user becoming a paying customer is predicted based on collected behavior data corresponding to each prospective user. A likelihood of an application corresponding to the service and associated with a particular user will misbehave is predicted based on resource usage and features of the application. Resource allocation to the application corresponding to the service and associated with the particular user is customized based the likelihood that the application will misbehave and whether the particular user is a current user or a prospective user.

First claim

What is claimed is:

1. A method for customizing computing resource allocation based on machine learning, the method comprising:
evaluating a plurality of current users of a service based on attributes of each respective current user in the plurality of current users;
generating a user value score for each respective current user in the plurality of current users based on the attributes of each respective current user;
clustering the plurality of current users into a plurality of user groups based on the user value score of each respective current user in the plurality of current users;
predicting, using the machine learning, a likelihood of each prospective user in a plurality of prospective users becoming a paying customer based on collected behavior data corresponding to each prospective user;
predicting, using the machine learning, a likelihood of an application corresponding to the service and associated with a particular user will misbehave based on resource usage of the application and features of the application;
customizing resource allocation to the application corresponding to the service and associated with the particular user based on the likelihood that the application will misbehave and whether the particular user is a current user or a prospective user; and
selecting the particular user from one of the plurality of current users and the plurality of prospective users, wherein the customized resource allocation represents a customized allocation of cloud resources to the application, and the cloud resources comprises at least one processor resource, memory resources, storage resources and network resources, wherein the user value score for each respective current user represents a degree of importance the each respective current user is to the service that the application corresponds to, wherein when the particular user is a prospective user, the customizing of the resource allocation to the application is further based on the likelihood that the prospective user will convert to the paying customer, wherein the resource usage of the application is selected from a group consisting of processor usage, memory usage, storage usage, and network usage, and further comprising:
identifying, using the machine learning, a classification for the application based on the resource usage of the application and the features of the application, wherein the classification represents an application class, category or type corresponding to the application.

2. The method of claim 1 , wherein when the particular user is a current user, the customizing of the resource allocation to the application is further based on a user group of the plurality of user groups corresponding to the particular user.

3. The method of claim 1 , wherein the generated user value score is calculated using the machine learning, and wherein the attributes of each respective current user are selected from a group consisting of user recency data representing how recently has the each respective current user utilized resources of a cloud service, user frequency data representing how often the each respective current user utilizes the resources of the cloud service, and user revenue data that represents how much the each respective current user pays for utilizing the resources of the cloud service.

4. The method of claim 1 , wherein the collected behavior data corresponding to each prospective user that is the basis for predicting the likelihood of each respective user of the plurality of users becoming the paying customer include a history of which applications a particular prospective user used during a trial period, a frequency of application use, and how long an application was used during a trial session of the prospective user.

5. The method of claim 1 , wherein the application corresponding to the service is classified as one of a trial application of a non-paying subscription to a cloud service, or a zombie application that is inactive but continues to consume the cloud resources.

6. The method of claim 1 , wherein the service is a cloud service, and further comprising:
placing each application in a set of applications of the cloud service in a particular application classification of application classifications in a matrix;
placing each prospective user in a particular customer classification of customer classifications in the matrix; and
utilizing the matrix to determine how to customize the resource allocation to the application for each particular user of the cloud service based on the application classification and the customer classification corresponding to the particular user.

7. A computer system for customizing computing resource allocation based on machine learning, the computer system comprising:
a bus system;
a storage device connected to the bus system, wherein the storage device stores program instructions; and
a processor connected to the bus system, wherein the processor executes the program instructions to:
evaluate a plurality of current users of a service based on attributes of each respective current user in the plurality of current users;
generate a user value score for each respective current user in the plurality of current users based on the attributes of each respective current user;
cluster the plurality of current users into a plurality of user groups based on the user value score of each respective current user in the plurality of current users;
predict, using the machine learning, a likelihood of each prospective user in a plurality of prospective users becoming a paying customer based on collected behavior data corresponding to each prospective user;
predict, using the machine learning, a likelihood of an application corresponding to the service and associated with a particular user will misbehave based on resource usage of the application and features of the application;
customize resource allocation to the application corresponding to the service and associated with the particular user based on the likelihood that the application will misbehave and whether the particular user is a current user or a prospective user; and
select the particular user from one of the plurality of current users and the plurality of prospective users, wherein the customized resource allocation represents a customized allocation of cloud resources to the application, and the cloud resources comprises at least one processor resource, memory resources, storage resources and network resources, wherein the user value score for each respective current user represents a degree of importance the each respective current user is to the service that the application corresponds to, wherein when the particular user is a prospective user, customizing of the resource allocation to the application is further based on the likelihood that the prospective user will convert to the paying customer, wherein the resource usage of the application is selected from a group consisting of processor usage, memory usage, storage usage, and network usage, and wherein the processor further executes the program instructions to:
identify a classification for the application based on the resource usage of the application and the features of the application, wherein the classification represents an application class, category or type corresponding to the application.

8. The computer system of claim 7 , wherein when the particular user is a current user, customizing the resource allocation to the application is further based on a user group of the plurality of user groups corresponding to the particular user.

9. A computer program product for customizing computing resource allocation based on machine learning, the computer program product comprising a computer readable storage med

Assignees

• IBM

Inventors

• Zhang Bo
• Higgins Chunhui Y
• Ran Chuan
• Schipon Nancy Anne
• Cui Yongchang

Classifications

• G06N5/01

  Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound · CPC title

• G06Q10/06315 Primary

  Needs-based resource requirements planning or analysis · CPC title

• G06N20/00

  Machine learning · CPC title

• G06N5/048

  Fuzzy inferencing · CPC title