dataflow ML. Can't Apply a Trained Model to a New Dataset

[olegkazanskyi]

I taught a ML classification model in the dataflow.

That was easy and it's great.

Now I want to add a new table to the dataflow with the same columns except target one, to get predictions based on a model.

But I can't do this because dataflow shows errors in edit mode and I can't understand the type of errors.

The errors are in the model folder (I named a model "1st try").

It seems it asks me to fill parameters, but I can't find any tutorial about this.

Any suggestions?

[olegkazanskyi]

Hello @v-luwang-msft ,

Thank you for the idea of where the root might be.

But dataflow doesn't have a DataSource setting button.

In the "ProjectOption" I enabled combining data from many sources.

But it did not change anything. 

Am I doing something wrong?

[v-luwang-msft]

Hi @olegkazanskyi ，

make sure that the privacy level for both data sources is compatible (Public/Public or Organizational/Organizational)，or disable the firewall through the "Fast Combine" option. 

 I think here could be the solution. Please give it a try.

Did I answer your question? Mark my post as a solution!

Best Regards

Lucien