12.2 Create a HeatWave AutoML model

Use the HeatWave Console to create a new HeatWave AutoML model.

Create a HeatWave AutoML model includes an advanced option to choose one or more algorithms. HeatWave AutoML uses the selected algorithms as a pool of algorithms, and will choose the most applicable algorithm from this pool to evaluate the model. By default, all algorithms are selected. To use a single, specific algorithm to evaluate the model, deselect all the other algorithms.

1. Connect to the DB System. See Connecting from the Console.

2. Select the Workspace tab in the HeatWave Console, and then select the Model Catalog tab and click Create Model

3. In the Create Machine Learning Model pane that opens, provide the following information:

   • Name: Specify a name for the model or use the generated default name.
   • Description: Specify a user friendly description for the model.
   • Training table: Select a training dataset.
   • Columns to include in the model: Select the Target column, and then choose which columns to Include in model.
   • Machine learning task: Choose a task.

   • Advanced:

     • Optimization metric: Choose an optimization metric.

     • Algorithms to evaluate: Choose which algorithms HeatWave AutoML should consider to evaluate the model.

4. Click Create. The HeatWave Console returns to the Model Catalog tab, and shows the new model.

Parent topic: HeatWave AutoML