AI Prediction Model: Result column is blank

[FlorentineBader]

Hi, I created an AI model to predict customer churn. The Model has an accuracy of 99% (level A). After publishing the model, the result columns have been created in my database. However, the result columns are empty:

Is there an explanation for this? Thanks in advance!

[AnnaChu]

Thanks for your question @FlorentineBader. You may want to join us for an 'Ask Microsoft Anything' (AMA) tomorrow to ask this question live! Join us 8:00AM Pacific Time (UTC-8) on Tuesday June 7 https://aka.ms/MPA_CommunityAMA where the AI Builder team will be available to answer your questions in real time.

[CedrickB]

You have 2 options, the first one is to use the scheduled prediction

See  https://docs.microsoft.com/ai-builder/prediction-use#schedule-retrain-and-run-preview 

Alternatively, you can leverage the real time prediction using the Power Automate action either by providing the Datatverse row id

or the list of fields

Then you can save the output in the appropriate column of Datatverse using the "update a row" action of Dataverse.

[FlorentineBader]

Hi Cedrick, 

thanks for your help!
I tried to use the real time prediction approach with Power Automate. 
However, I got the following error message for the "Update a row" step: 

The error said: "404 - File or directory not found. The resource you are looking for might have been removed, had its name changed, or is temporarily unavailable." 
It seems like the probability of the row was correctly predicted, but there was a problem with updating the row. Do you have any ideas why this is happening? As far as I can see, everything is fine with the table I want to update. 

Thank you!

[CedrickB]

Can you send a screenshot of the flow please?

This is a basic Dataverse table which should work fine.

The message is not super clear but it looks like the table you have entered doesn't exist anymore.

[FlorentineBader]

Sure! 

 

The trigger is "When record is updated", then I use my AI Model for the prediction and try to update the rows of the table in step 1. The table I am referring to is not the table I was using to train the model. But that should not be a problem right?

[CedrickB · Accepted solution]

For the rowid, you should past the Churn_Prediction_NS id instead of E-mail.

[FlorentineBader]

Perfect, it worked! Thank you 🙂

[FlorentineBader]

Do you know if there is a way to apply the prediction on the whole table and not just on one row when the row when gets updated? I guess this needs a new trigger. However, if I manually trigger I do not have the connection to the specific file. 
Thanks in advance!

[CedrickB · Accepted solution]

You can create a manually triggered or scheduled flow that will use Dataverse "List rows" then call the prediction and update the row in an "Apply to each" action. 

[Anonymous]

Hi guys, I have been tried using the sample model of Movie Analysis and been caught up with the same problem. 

Can I view the results without creating a flow? If not, can I have a sample of the power automate flow for this step?

[bisheshm]

I am having the same issue - shouldn't the prediction columns automatically be populated once you publish and run the model?

[Anonymous]

I have the SAME problem but neither of these solutions work. I have the license only of Power App and so I can't try the Power automate solution. Please help me!