Machine learning system for identifying potential escalation of customer service requests
A method, system and computer-usable medium are disclosed for machine learning to identify service request records associated with an account that is likely to escalate. Certain aspects of the disclosure include generating a random forest model using a training set of service request records to determine a probability of escalation for service requests of the training set; applying the random forest model to a current set of service request records to determine an escalation probability for service requests in the current set; and assigning service request records in the current set to a plurality of escalation probability bins, wherein the service request records of the current set are generally equally divided between the plurality of escalation probability bins, and wherein the service request records of the current set are assigned to a probability bin based on the escalation probability of the service request record.
1. A computer-implemented method for identifying potential escalations of customer service requests, comprising:
generating a random forest model using a training set of service request records to determine a probability of escalation for service requests of the training set, service request records of the training set including data relating to independent variables that are analyzed while generating the random forest model;
applying the random forest model to a current set of service request records to determine an escalation probability for service requests in the current set;
assigning service request records in the current set to a plurality of escalation probability bins, wherein the service request records of the current set are generally equally divided between the plurality of escalation probability bins, and wherein the service request records of the current set are assigned to a probability bin based on the escalation probability of the service request record
identifying a subset of independent variables during generation of the random forest model that are most influential in determining the escalation probabilities;
assigning service request records in a test set to a plurality of escalation probability bins, wherein the test set of service request records are generally equally divided between the plurality of escalation probability bins, wherein service request records of the test set are assigned to a probability bin based on the escalation probability of the service request record;
generating a decision tree using the subset of independent variables, the decision tree being generated using service request records in a probability bin associated with a highest escalation probability; and,
applying rules of the decision tree to current service request records assigned to the probability bin associated with the highest escalation probability to further identify whether one or more of the current service request records are likely to escalate.
2. The computer-implemented method of claim 1 , wherein
the plurality of escalation probability bins divide the service request records in the current set into deciles.
3. The computer-implemented method of claim 1 , further comprising:
under sampling the training set of service request records for use in generating the random forest model.
4. The computer-implemented method of claim 1 , wherein the service request records of the training set include data relating to independent variables that are analyzed while generating the random forest model, the method further comprising:
identifying a subset of independent variables for use in the random forest model using one or more of a mean decrease accuracy and/or mean decrease gini technique to identify independent variables that are most influential in determining the escalation probabilities; and
tuning the random forest model to limit the independent variables used in the random forest model to the subset of independent variables.
5. The computer-implemented method of claim 4 , wherein the independent variables include one or more of order level variables and transaction level variables.
6. A system comprising:
a processor;
a data bus coupled to the processor; and
a non-transitory, computer-readable storage medium embodying computer program code, the non-transitory, computer-readable storage medium being coupled to the data bus, the computer program code interacting with a plurality of computer operations and comprising instructions executable by the processor and configured for:
generating a random forest model using a training set of service request records to determine a probability of escalation for service requests of the training set, service request records of the training set including data relating to independent variables that are analyzed while generating the random forest model;
applying the random forest model to a current set of service request records to determine an escalation probability for service requests in the current set;
assigning service request records in the current set to a plurality of escalation probability bins, wherein the service request records of the current set are generally equally divided between the plurality of escalation probability bins, and wherein the service request records of the current set are assigned to a probability bin based on the escalation probability of the service request record;
identifying a subset of independent variables during generation of the random forest model that are most influential in determining the escalation probabilities;
assigning service request records in a test set to a plurality of escalation probability bins, wherein the test set of service request records are generally equally divided between the plurality of escalation probability bins, wherein service request records of the test set are assigned to a probability bin based on the escalation probability of the service request record;
generating a decision tree using the subset of independent variables, the decision tree being generated using service request records in a probability bin associated with a highest escalation probability; and,
applying rules of the decision tree to current service request records assigned to the probability bin associated with the highest escalation probability to further identify whether one or more of the current service request records are likely to escalate.
7. The system of claim 6 , wherein
the plurality of escalation probability bins divide the service request records in the current set into deciles.
8. The system of claim 6 , wherein the instructions are further configured for:
under sampling the training set of service request records for use in generating the random forest model.
9. The system of claim 6 , wherein the service request records of the training set include data relating to independent variables that are analyzed while generating the random forest model, wherein the instructions are further configured for:
identifying a subset of independent variables for use in the random forest model using one or more of a mean decrease accuracy and/or mean decrease gini technique to identify independent variables that are most influential in determining the escalation probabilities; and
tuning the random forest model to limit the independent variables used in the random forest model to the subset of independent variables.
10. The system of claim 9 , wherein the independent variables include one or more of order level variables and transaction level variables.
11. A non-transitory, computer-readable storage medium embodying computer program code, the computer program code comprising computer executable instructions configured for:
generating a random forest model using a training set of service request records to determine a probability of escalation for service requests of the training set, service request records of the training set including data relating to independent variables that are analyzed while generating the random forest model;
applying the random forest model to a current set of service request records to determine an escalation probability for service requests in the current set;
assigning service request records in the current set to a plurality of escalation probability bins, wherein the service request records of the current set are generally equally divided between the plurality of escalation probability bins, and wherein the service request records of the current set are assigned to a probability bin based on the escalation probability of the service request record;
identifying a subset of independent variables during generation of the random forest model that are most influential in determining the escalation probabilities;
assigning service request records in a test set to a plurality of escalation probability bins, wherein the test set of service request records are generally equally divided between the plurality of escalation probability bins, wherein service request records of the test set are assigned to a probability bin based on the escalation probability of the service request record;
generating a decision tree using the subset of independent variables, the decision tree being generated using service request records in a probability bin associated with a highest escalation probability; and,
applying rules of the decision tree to current service request records assigned to the probability bin associated with the highest escalation probability to further identify whether one or more of the current service request records are likely to escalate.
12. The non-transitory, computer-readable storage medium of claim 11 , wherein
the plurality of escalation probability bins divide the service request records in the current set into deciles.
13. The non-transitory, computer-readable storage medium of claim 11 , wherein the instructions are further configured for:
under sampling the training set of service request records for use in generating the random forest model.
14. The non-transitory, computer-readable storage medium of claim 11 , wherein the service request records of the training set include data relating to independent variables that are analyzed while generating the random forest model, wherein the instructions are further configured for:
identifying a subset of independent variables for use in the random forest model using one or more of a mean decrease accuracy and/or mean decrease gini technique to identify independent variables that are most influential in determining the escalation probabilities; and
tuning the random forest model to limit the independent variables used in the random forest model to the subset of independent variables.