Creating Experiment Model

Create algorithm models out of the published experiments.

1. Log in to supOS, and then select Data Intelligence > Data Science > Model Management.
2. Click Add > Add experimental model, and then enter the basic information of the model.
   info

   Click Import, and then select third-party models to upload to X-BD. For developing details, see Developing Third-party Algorithm Model.

Parameter	Description
Related experiments	Select an experiment that is successfully debugged to be the base of the model.
Code	Automatically generated. You can manually change it. It must be unique.
Support update	Enable it, and when the requirements are met, the model can be updated through openAPI or X-ETL.
Benchmark model	Select a component that is categorized as Model Assessment in the experiment.
Health threshold	When the debug result is higher than the set health threshold, the model is updated. The threshold needs to be within 60-100.
Batch debug times	When debugging the experiment, click Batch debug, the experiment will be debugged the set times. The number must be within 1-100.

3. Click Next, and then edit the input/output fields of the experiment to be the input/output fields of the model.

   tip

   Click Preview to see the effect.

   

4. Click Submit.