Import models to the model repository

The model repository displays status details for all the imported and deployed models across the platform. Learn how to add and delete models to the repository.

In This Topic

Import a model
Deploy from the model repository
Delete a model

Import a model

Importing a model is like opening a file.

Sign in to Minitab Model Ops to open the Model Repository .
Select Import a model.
In Project File, browse to and select the file that contains a model to import.
If you import a Minitab project file that contains multiple models, select the model from the dropdown list which contains all the supported models in the file.
Note
For more information on supported models, go to Supported model types.
Select Use training data for baseline data to upload the training data with the model. For best results, it is good practice to upload the training data as the initial baseline data. If your training data has changed or you don't want to use it as baseline data, you can upload baseline data later. For more information, go to Baseline data.
In Model Name, enter a descriptive name for the file.
Select Import Model.

When you import a binary logistic model from Minitab Statistical Software, Minitab Model Ops requires the binary event probability threshold. In Binary Event Probability Threshold, enter a value between 0 and 1. The model assigns cases with an event probability greater than the threshold to the event class. For example, a value of 0.5 means that the model assigns a case to the event class when the probability of the event is higher than 0.5.

Note

The model upload fails when models have more than 500 predictors or for a multinomial response variable, have more than 10 classes. You can check the System Audit Log for more information.

Deploy from the model repository

After you import a model to the Model Repository, the status is Ready. To access a model from its endpoint and make predictions, create a deployment that contains one champion and up to two challenger models.

Sign in to Minitab Model Ops to open the Deployments page.
Select New Deployment.
In Deployment Name, enter a descriptive name for the deployment.
Select Select From Repository then Next.
In Model, select an initial champion model. You can select only a model that is Ready and not used in another deployment. The list of models is in alphabetical order by name, with uppercase letters before lowercase letters.
Select Create Deployment.

Delete a model

You can delete an imported model from the Model Repository page. Only models that have Ready status can be deleted. If a model is already deployed, you must first remove it from the deployment before you can delete it.

Select the Delete button to delete a model. When a model is deleted, all historical data are also removed.