Creating Custom Project Groups to Train Models

To ensure models are trained to predict outcomes with a high level of accuracy, you can create custom project groups, by selecting a subset of similar projects that have been completed in the past to train your ML models with your data.

To set up custom project groups to train models with your data:

Sign in to the CIC Administration application.
http://<host>:<port>/cicadmin
From the Home menu, select a staretl<id> or a <Custom Name for DataSource>.
Note: <id> is the identifier of the data source.
In the sidebar, click Custom Project Groups to create a maximum of three groups.
On the Custom Project Groups page, click Edit to rename a project group.
1. To add a project to a group, click and drag a project to the relevant project group column..
2. To remove a specific project a project from a group, select the project in the project group and click Remove.
3. Repeat the above actions as necessary to create each combination of projects in a group.
Click Save to confirm all changes or click Cancel to revert all changes made to the project groups.
Proceed to select any of the custom project groups created above and train a model with your data on the ML Workbench page.

Related Topics