How to evaluate the performance of supervised learning of auto model？

AP

How to evaluate the performance of supervised learning of auto model？Is there any documentation about this？

Caperez

Hi @AP, 

Into Automodel you have a button "Open Process" click on and then you can see the scoring, validation block.
Inside you have the validation subprocess.

Best, 

Cesar

  

Caperez

Hi @AP,

You have two ways.
1- In each model used in Automode you have a specifif topic for performance with some performance meassurements.
2- You can open the automodel process and then, into the performance opertator, change the automatically selected metrics for your desired ones.

Best

Cesar

AP

What I want to know is Whether it's based on training set or test set

Caperez

Hi again @AP, 

when you open the model, the performance operator is applied to validation set.

Best.

Cesar

AP

The performance is based on cross validation？default ten fold？

Caperez

Hi @AP, 

when you open the proposed model in automodel you can check the validation process. 

As far I remember the system use to work with a Hold-out set but I don't remeber the number of partitions. 

You can check it in your results. 

AP

“when you open the proposed model in automodel you can check the validation process. ”--How can I do it？Can you give me a graph？

Caperez

Hi @AP, 

Into Automodel you have a button "Open Process" click on and then you can see the scoring, validation block.
Inside you have the validation subprocess.

Best, 

Cesar

  

AP

Yes, I see. Thank you very much