Hi All
I created an AI model using Prediction (Binary Classification) in February and it worked like a charm. That was my trial, it was expired!
Today, I tried the same dataset and created the model but this time, it never predicts (even after re-training, re-publishing) or clicking on Use Model and Run.
I thought there must be something wrong with my technique, I followed the guide provided in docs related to the aib_onlineshopperintention - Unfortunately all my efforts have gone in vain.
My findings
My label (like online shopper model) is of Two Option type. Therefore, whenever I create a new record, it sets as No (because it must have a default value). I got to know earlier (in January) that you must have your label set as null in order for model to generate the prediction in the new (system created) field. That worked fine then (till February).
Also, earlier, there used to be three new fields created by the AI Builder (prediction, probably and explanation).
However, this time, neither Online Shopper nor my model work with this trick.
What could go wrong? Is it something I am not following or there's a genuine issue?
Looking forward to hear from you.
Thanks
Arafat
I have faced similar issue & the below workaround might work: Our data has only int datatype (whole number) columns. When AI Builder predicts new predictions, it gets float values as well. But it couldn't display them for some reason. So the solution for this would be to change the datatype of target column to float while loading the data into the Dataverse table and this way AI Builder displays the predicted values as float values.
Can you please take a look at this post please
Re: AI Prediction Model: Result column is blank - Power Platform Community (microsoft.com)
Hi @Norah / @ArafatTehsin ,
Could you please let me know how the issue has been resolved. I am facing similar issue with the model not generating predictions for regression problem. Any help on this would be appreciated. Thank you.
hi Arafat
Let me contact you via private message and get some more details.
Thanks
Norah