Machine Learning Designer provides preset pipeline templates that can be used to quickly create pipelines. If you want to create a pipeline that is significantly different from any preset template, you can create a blank pipeline and add components to the pipeline to build models. The topic describes how to create a blank pipeline.
Prerequisites
A workspace is created. For more information, see Create a workspace.
Procedure
Go to the Machine Learning Designer page.
Log on to the PAI console.
In the left-side navigation pane, click Workspaces. On the Workspaces page, click the name of the workspace that you want to manage.
In the left-side navigation pane, choose .
On the Pipelines tab, choose .
In the Create Pipeline dialog box, configure the parameters that are described in the following table and click OK.
Parameter
Description
Pipeline Name
The name of the pipeline. Specify a name as prompted.
Pipeline Data Path
The path of the Object Storage Service (OSS) bucket that stores the temporary data and models generated during pipeline runtime. We recommend that you set this parameter.
The system automatically creates a temporary directory in the
<Pipeline data path>/<Task ID>/<Node ID>format for each run. This saves you from creating an OSS directory for storing data of each component and allows you to manage data in a centralized manner.Description
The description of the pipeline. Descriptions are used to identify pipelines.
Visibility
The scope at which the pipeline is visible. Valid values:
Visible to Me: The pipeline is created under the My Pipelines folder and is visible only to you and the administrators of the current workspace.
Visible to Current Workspace: The pipeline is created under the Pipelines Visible to Workspaces folder and is visible to all members of the current workspace.
What to do next
After you create a pipeline, you can go to the configuration tab of the pipeline and build a model with ease in a visualized manner. For more information, see Model training.