Method and system for predicting user involvement requirements for upgrade failures
A method for managing upgrades of components of clients includes obtaining an upgrade failure prediction request associated with a client of the clients, and in response to obtaining an update failure prediction request: obtaining live data associated with the client, matching the live data with a training data cluster, selecting relevant features associated with processed training data of the training data cluster, generating an upgrade failure prediction using the live data associated with the relevant features and a prediction model, making a determination that the upgrade failure prediction implicates an action is required, and based on the determination, initiating performance of the action.
1. A method for managing upgrades of components of clients, comprising:
obtaining an upgrade failure prediction request associated with an upgrade of an upgrade type of upgrade types, wherein the upgrade types comprise:
a software upgrade of a client software component of a client of clients, and
a hardware upgrade of a client hardware component of the client;
in response to obtaining an update failure prediction request:
obtaining live data associated with the client;
matching the live data with a training data cluster;
selecting relevant features associated with processed training data of the training data cluster using a feature selection algorithm and the training data cluster, wherein:
the processed training data comprises a plurality of features comprising client configuration information features, client upgrade history information features, and upgrade failure history information features;
the feature selection algorithm selects and specifies the relevant features associated with the training data cluster based on Z-scores associated with the plurality of features; and
the relevant features comprise a portion of the client configuration information features, a portion of the client upgrade history information features, and a portion of the upgrade failure history information features in the processed training data that are highly relevant to upgrade failure predictions;
generating an upgrade failure prediction using the live data associated with the relevant features and a prediction model;
making a determination that the upgrade failure prediction implicates an action is required; and
based on the determination:
initiating performance of the action.
2. The method of claim 1 , further comprising:
before obtaining the update failure prediction request:
identifying a training event associated with the clients;
in response to identifying the training event:
obtaining raw training data associated with the clients;
obtaining processed training data using the raw training data;
generating clustered training data using the processed training data; and
generating a prediction model by applying the processed training data to a classification algorithm.
3. The method of claim 2 , wherein the raw training data comprise training data obtained from at least two of the clients.
4. The method of claim 2 , wherein the live data comprises second training data obtained from only the client.
5. The method of claim 2 , wherein the raw training data comprises at least one selected from a group consisted of:
the client configuration information features,
the client upgrade history information features, and
the upgrade failure history information features.
6. The method of claim 5 , wherein generating clustered training data comprises separating the processed training data into training data clusters, wherein a training data cluster of the training data clusters comprises at least one selected from a group consisting of:
processed training data that comprise at least a portion of the same client configuration information features,
processed training data that comprise at least a portion of the same client upgrade history information features, and
processed training data that comprise at least a portion of the same upgrade failure history information features.
7. The method of claim 2 , wherein the raw training data comprise more features than the relevant features.
8. The method of claim 1 , wherein the action comprises notifying a user that user intervention will be required to resolve the upgrade failure.
9. A system for managing upgrades of components of clients comprises:
persistent storage for storing:
raw training data, and
processed training data; and
a recommendation system programmed to:
obtain an upgrade failure prediction request associated with an upgrade of an upgrade type of upgrade types, wherein the upgrade types comprise:
a software upgrade of a client software component of a client of clients, and
a hardware upgrade of a client hardware component of the client;
in response to obtaining an update failure prediction request:
obtain live data associated with the client;
match the live data with a training data cluster;
select relevant features associated with processed training data of the training data cluster using a feature selection algorithm and the training data cluster, wherein:
the processed training data comprises a plurality of features comprising client configuration information features, client upgrade history information features, and upgrade failure history information features;
the feature selection algorithm selects and specifies the relevant features associated with the training data cluster based on Z-scores associated with the plurality of features; and
the relevant features comprise a portion of the client configuration information features, a portion of the client upgrade history information features, and a portion of the upgrade failure history information features in the processed training data that are highly relevant to upgrade failure predictions;
generate an upgrade failure prediction using the live data associated with the relevant features and a prediction model;
make a determination that the upgrade failure prediction implicates an action is required; and
based on the determination:
initiate performance of the action.
10. The system of claim 9 , wherein the recommendation system is further programmed to:
before obtaining the update failure prediction request:
identify a training event associated with the clients;
in response to identifying the training event:
obtain raw training data associated with the clients;
obtain processed training data using the raw training data;
generate clustered training data using the processed training data; and
generate a prediction model by applying the processed training data to a classification algorithm.
11. The system of claim 10 , wherein the raw training data comprise training data obtained from at least two of the clients.
12. The system of claim 10 , wherein the live data comprise second training data obtained from only the client.
13. The system of claim 10 , wherein the raw training data comprise at least one selected from a group consisted of:
the client configuration information features,
the client upgrade history information features, and
the upgrade failure history information features.
14. The system of claim 13 , wherein generating the clustered training data comprises separating the processed training data into training data clusters, wherein a training data cluster of the training data clusters comprises at least one selected from a group consisting of:
processed training data that comprise at least a portion of the same client configuration information features,
processed training data that comprise at least a portion of the same client upgrade history information features, and
processed training data that comprise at least a portion of the same upgrade failure history information features.
15. The system of claim 10 , wherein the raw training data comprise more features than the relevant features.
16. The system of claim 9 , wherein the action comprises notifying a user that user intervention will be required to resolve the upgrade failure.
17. A non-transitory computer readable medium comprising computer readable program code, which when executed by a computer processor enables the computer processor to perform a method for managing upgrades of components of clients, the method comprising:
obtaining an upgrade failure prediction request associated with an upgrade of an upgrade type of upgrade types, wherein the upgrade types comprise:
a software upgrade of a client software component of a client of clients, and
a hardware upgrade of a client hardware component of the client;
in response to obtaining an update failure prediction request:
obtaining live data associated with the client;
matching the live data with a training data cluster;
selecting relevant features associated with processed training data of the training data cluster using a feature selection algorithm and the training data cluster, wherein:
the processed training data comprises a plurality of features comprising client configuration information features, client upgrade history information features, and upgrade failure history information features;
the feature selection algorithm selects and specifies the relevant features associated with the training data cluster based on Z-scores associated with the plurality of features; and
the relevant features comprise a portion of the client configuration information features, a portion of the client upgrade history information features, and a portion of the upgrade failure history information features in the processed training data that are highly relevant to upgrade failure predictions;
generating an upgrade failure prediction using the live data associated with the relevant features and a prediction model;
making a determination that the upgrade failure prediction implicates an action is required; and
based on the determination:
initiating performance of the action, wherein the action comprises notifying a user that user intervention will be required to resolve the upgrade failure.
18. The non-transitory computer readable medium of claim 17 , wherein the method further comprises:
before obtaining the update failure prediction request:
identifying a training event associated with the clients;
in response to identifying the training event:
obtaining raw training data associated with the clients, wherein the raw training data comprise at least one selected from a group consisted of: client configuration information features, client upgrade history information features, and upgrade failure history information features;
obtaining processed training data using the raw training data;
generating clustered training data using the processed training data; and
generating a prediction model by applying the processed training data to a classification algorithm.
19. The non-transitory computer readable medium of claim 18 , wherein the raw training data comprise training data obtained from at least two of the clients.
20. The non-transitory computer readable medium of claim 18 , wherein generating the clustered training data comprises separating the processed training data into training data clusters, wherein a training data cluster of the training data clusters comprises at least one selected from a group consisting of:
processed training data that comprise at least a portion of the same client configuration information features,
processed training data that comprise at least a portion of the same client upgrade history information features, and
processed training data that comprise at least a portion of the same upgrade failure history information features.