To facilitate the key features mentioned in the previous section, teradataml AutoML provides the following methods:
- _init_(): Create instance for AutoML training.The following APIs are available for instance creation based on problem types:
- AutoML (AutoML.__init__): Generic AutoML API that can be used to solve all three supported problem type.
- AutoRegressor (AutoRegressor.__init__): Specific AutoML API to solve regression problem type.
- AutoClassifier (AutoClassifier.__init__): Specific AutoML API to solve classification problem type.
- AutoFraud AutoFraud.__init__: Specific AutoML API to solve fraud detection problem type.
- AutoChurn AutoChurn.__init__: Specific AutoML API to solve churn prediction problem type.
- AutoCluster AutoCluster.__init__: Specific AutoML API to solve clustering task problem type.
The following methods are common for these APIs (AutoML, AutoRegressor, AutoClassifer, AutoFraud, AutoChurn, and AutoCluster).load(), deploy(), evaluate(), and visualize() methods are not applicable for AutoCluster.
- fit: Fit over given dataset during AutoML training.
- leaderboard: Display model leaderboard containing model rank and corresponding performance metrics.
- leader: Display best performing model.
- model_hyperparameters: Get hyperparameters of the model based on rank in leaderboard.
- predict: Generate prediction and performance metrics.
- evaluate: Evaluate on data using model rank in leaderboard to generate performance metrics.
- generate_custom_config: Generate custom config JSON required for customized run.
- get_persisted_tables: List the persisted tables created during execution.
- deploy: Save models to the specified table.
- load: Load models information from the specified table.
- remove_saved_models: Remove the specified table containing saved models.
- visualize: Visualize the teradataml DataFrame.