predictive intelligent change capability when train

[Cherly]

After the upgrade to Vancouver. when we train the capability change to from classification to Workflow Classification. is there any way to change fix this issue.

[k_lutz]

Thanks Brian. It was great meeting up and getting more insights into the platform. I do have a follow-up question now that I have thought about it some more. You mentioned you can train a solution on a custom table and then call it via the business rule to be applied to another table...but how does that work for it to know the output field? Does it just need to reference the same table on the custom table? Not sure I understand how it would understand the data from the custom table and apply the outcome to Incident.

[Brian Bakker]

@k_lutz 

When you train the solution on a custom table, instead of the base table, such as [incident], as long as the same Input and Output fields exist on both tables, you can either Use Predictive Intelligence in Flow Designer with ML actions to create a trigger on the [incident] table to make a prediction on the trained solution used on the custom table or you can use a Before Insert BR on table [incident] and trigger a prediction on the trained solution used on the custom table by Using Machine Learning APIs. In both cases, you reference the "Solution name" of the trained solution in table [ml_solution] that used the custom table as the Input table, and the flow trigger or the Before Insert BR on table [incident] will ensure the prediction on the solution trained on the custom table is called on the [incident] table.

I hope this clarifies it.

Regards,

Brian

[k_lutz]

Hi Brian,

Just a clarification on what you mentioned abut same fields. If I was using incident, the field would be short_description. If I am using a custom table, then my field would be u_short_description which would not be the same or is it the label needs to be the same?

Thanks.

[Brian Bakker]

@k_lutz 

Good question... and I will find the answer, so please bear with me.

[k_lutz]

Thanks Brian. I am checking with our architect to see if he would be OK with me creating an extended table from task but locking it down (either with ACLs or data filtration) to the admin / ML roles as a way to have it with the same names but protect it from anyone else.