Method and system for predicting next steps for customer support cases
Techniques described herein relate to a method for managing support cases of clients. The method may include obtaining a support case associated with a client of the clients; in response to obtaining the support case, obtaining contextual information associated with the support case; obtaining client telemetry data associated with the support case; selecting a portion of training data that is similar to the support case based on the client telemetry data, the contextual information, and the support case; generating a prediction model using the portion of the training data and a classification algorithm; and generating predictions using the prediction model, the client telemetry data, the contextual information, and the support case.
1. A method for managing support cases associated with clients, comprising:
obtaining, by a computing device, a support case associated with a failure of a client of the clients;
in response to obtaining the support case:
obtaining, by the computing device, contextual information associated with the support case, wherein the contextual information comprises:
client component identifiers,
client component types,
client component performance information,
temperature information, and
issue severity;
obtaining, by the computing device, client telemetry data associated with the support case;
selecting, by the computing device, a portion of training data that is similar to the support case based on a similarity calculation between the client telemetry data, the contextual information, and the support case and the training data, wherein the training data comprises past contextual information from previous support cases of the clients, wherein the portion of the training data comprises baseline training data stored on a training data repository;
generating, by the computing device, a prediction model by executing a classification algorithm using the portion of the training data, wherein the prediction model reflects a subset of features included in the portion of the training data; and
generating, by the computing device, predictions by executing the prediction model using the client telemetry data, the contextual information, and the support case, wherein generating the predictions using the client telemetry data, the contextual information, and the support case comprises:
identifying a number of times each prediction of the predictions occurs in the predictions, and
ranking the predictions of the predictions based on the number of times each prediction occurs to generate ranked predictions, wherein a highest occurring prediction is ranked highest.
2. The method of claim 1 , wherein the predictions comprise steps to be taken by a user to resolve the support case.
3. The method of claim 1 , wherein selecting the portion of the training data comprises:
generating similarity scores between previous support cases included in the training data and the support case; and
selecting the portion of the training data, wherein the portion of the training data comprises support cases associated with similarity scores above a threshold similarity score.
4. The method of claim 1 , wherein generating predictions using the prediction model, the client telemetry data, the contextual information, and the support case further comprises:
generating the predictions by applying the prediction model to the client telemetry data, the contextual information, and the support case;
providing the ranked predictions to a user via a graphical user interface;
obtaining a portion of the ranked predictions, wherein the portion of the ranked predictions were used by the user to resolve the support case; and
updating a cyclic training data of the training data repository to include the support case, the contextual information, the client telemetry data, and the portion of the predictions, wherein the updated cyclic training data is used to train new prediction models for new support cases.
5. The method of claim 4 , wherein the portion of the ranked predictions comprise the portion of the predictions that were used by the user to resolve the support case.
6. The method of claim 4 , wherein the cyclic training data comprises more data than the training data.
7. 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 support cases associated with clients, wherein the method comprising:
obtaining, by a computing device, a support case associated with a failure of a client of the clients;
in response to obtaining the support case:
obtaining, by the computing device, contextual information associated with the support case, wherein the contextual information comprises:
client component identifiers,
client component types,
client component performance information,
temperature information, and
issue severity;
obtaining, by the computing device, client telemetry data associated with the support case;
selecting, by the computing device, a portion of training data that is similar to the support case based on a similarity calculation between the client telemetry data, the contextual information, and the support case and the training data, wherein the training data comprises past contextual information from previous support cases of the clients, wherein the portion of the training data comprises baseline training data stored on a training data repository;
generating, by the computing device, a prediction model by executing a classification algorithm using the portion of the training data, wherein the prediction model reflects a subset of features included in the portion of the training data; and
generating, by the computing device, predictions by executing the prediction model using the client telemetry data, the contextual information, and the support case, wherein generating the predictions using the client telemetry data, the contextual information, and the support case comprises:
identifying a number of times each prediction of the predictions occurs in the predictions, and
ranking the predictions of the predictions based on the number of times each prediction occurs to generate ranked predictions, wherein a highest occurring prediction is ranked highest.
8. The non-transitory computer readable medium of claim 7 , wherein the predictions comprise steps to be taken by a user to resolve the support case.
9. The non-transitory computer readable medium of claim 7 , wherein selecting the portion of the training data comprises:
generating similarity scores between previous support cases included in the training data and the support case; and
selecting the portion of the training data, wherein the portion of the training data comprises support cases associated with similarity scores above a threshold similarity score.
10. The non-transitory computer readable medium of claim 7 , wherein generating predictions using the prediction model, the client telemetry data, the contextual information, and the support case further comprises:
generating the predictions by applying the prediction model to the client telemetry data, the contextual information, and the support case;
providing the ranked predictions to a user via a graphical user interface;
obtaining a portion of the ranked predictions, wherein the portion of the ranked predictions were used by the user to resolve the support case; and
updating a cyclic training data of the training data repository to include the support case, the contextual information, the client telemetry data, and the portion of the predictions, wherein the updated cyclic training data is used to train new prediction models for new support cases.
11. The non-transitory computer readable medium of claim 10 , wherein the portion of the ranked predictions comprise the portion of the predictions that were used by the user to resolve the support case.
12. The non-transitory computer readable medium of claim 10 , wherein the cyclic training data comprises more data than the training data.
13. A system for managing support cases associated with clients, the system comprising:
a persistent storage comprising a training data repository;
a computing device, comprising a processor and memory, configured to:
obtain a support case associated with a failure of a client of the clients;
in response to obtaining the support case:
obtain contextual information associated with the support case, wherein the contextual information comprises:
client component identifiers,
client component types,
client component performance information,
temperature information, and
issue severity;
obtain client telemetry data associated with the support case;
select a portion of the training data that is similar to the support case based on a similarity calculation between the client telemetry data, the contextual information, and the support case and the training data, wherein the training data comprises past contextual information from previous support cases of the clients, wherein the portion of the training data comprises baseline training data stored on a training data repository;
generate a prediction model by executing a classification algorithm using the portion of the training data, wherein the prediction model reflects a subset of features included in the portion of the training data; and
generate predictions by executing the prediction model using the client telemetry data, the contextual information, and the support case, wherein generating the predictions using the client telemetry data, the contextual information, and the support case comprises:
identifying a number of times each prediction of the predictions occurs in the predictions, and
ranking the predictions of the predictions based on the number of times each prediction occurs to generate ranked predictions, wherein a highest occurring prediction is ranked highest.
14. The system of claim 13 , wherein the predictions comprise steps to be taken by a user to resolve the support case.
15. The system of claim 13 , wherein selecting the portion of the training data comprises:
generating similarity scores between previous support cases included in the training data and the support case; and
selecting the portion of the training data, wherein the portion of the training data comprises support cases associated with similarity scores above a threshold similarity score.
16. The system of claim 13 , wherein generating predictions using the prediction model, the client telemetry data, the contextual information, and the support case comprises:
generating the predictions by applying the prediction model to the client telemetry data, the contextual information, and the support case;
providing the ranked predictions to a user via a graphical user interface;
obtaining a portion of the ranked predictions, wherein the portion of the ranked predictions were used by the user to resolve the support case; and
updating a cyclic training data of the training data repository to include the support case, the contextual information, the client telemetry data, and the portion of the predictions, wherein the updated cyclic training data is used to train new prediction models for new support cases.
17. The system of claim 16 , wherein the portion of the ranked predictions comprise the portion of the predictions that were used by the user to resolve the support case.
18. The system of claim 16 , wherein the cyclic training data comprises more data than the training data.