automl-pipeline-setup

Automating ML pipeline setup involves configuring data validation, feature engineering, AutoML training, and experiment tracking. This Skill uses YAML configuration to define data sources and targets, orchestrating the entire workflow for tabular data tasks.

Yes, you can use MLflow and H2O.ai together for automated model training and tracking. H2O.ai handles automated model selection and hyperparameter tuning, while MLflow logs experiments and manages the model registry throughout the pipeline execution.

Adding data validation to a machine learning pipeline is best achieved using Great Expectations. This integration performs robust data quality checks on your tabular data before feature engineering and model training occur, ensuring dataset integrity.

Yes, this automated ML pipeline supports orchestration with Apache Airflow. It includes an example Airflow DAG for pipeline scheduling, allowing you to automate the execution of data validation, training, and tracking on a defined schedule.

To configure a customer churn prediction pipeline, you define your data source and target variable in a YAML configuration file. The pipeline then automatically processes the tabular data, applies feature engineering, and trains the model.

You should not use an automated AutoML pipeline for non-tabular data like images or text, as it specifically supports tabular data for classification and regression. Complex, highly custom feature engineering requirements may also exceed its pre-built transformers.