How the Machine Learning mApp® Solution Works
The Machine Learning mApp Solution contains a machine learning model along with supporting expressions, fields, and a scheduled item designed to predict an Incident's Service Classification based on its description.
This mApp Solution covers this primary use case:
- Allow organizations to begin automating Incident workflows by having a Service Classification (Service, Category, and Subcategory) automatically assigned based on keywords entered into the Incident's description field.
When users or customers enter an Incident description, the machine learning model analyzes the entry and predicts a service, category, and subcategory, which gives the Service Classification. The prediction is assigned a confidence level. Machine learning subject matter experts (SMEs) can review the Service Classification predictions and verify the prediction or provide an updated classification. After a SME confirms or reclassifies, the prediction confidence is set to 100%. The machine learning model will train on all Service Classifications with a confidence of 95% or higher.
You may prefer to adjust how the machine learning model trains. Some ideas to consider:
- Exclude Service Classifications that have not been verified.
- Adjust the confidence threshold for training.
This mApp Solution includes:
- A machine learning model.
- A scheduled item that will run at certain intervals to perform the training.
- A supplemental Incident form containing fields for use in the main Incident form.
- One-Step™ Actions associated with the form fields.
- A dashboard the machine learning SMEs will use to verify the Service Classification predictions.
A separate .mApp file with training data is also available (machine-learning-data.mApp). Apply the data .mApp file after the main Machine Learning mApp Solution. Only install the data .mApp file on a separate demonstration environment, not your production system. Use your own data to train your production system.
For a list of items included in the Machine Learning mApp Solution, see Machine Learning mApp Solution Items.