In this comprehensive guide, we’ll dive deep into setting up Azure Machine Learning’s Automated ML for time-series forecasting. Unlike classical time series methods, Automated ML takes a unique approach by “pivoting” past time-series values to become additional dimensions for the regressor, along with other predictors. This approach allows Automated ML to incorporate multiple contextual variables and their relationships during training, making it a powerful tool for real-world forecasting scenarios.

We’ll start by preparing your data for time-series modeling, ensuring that all features used in training can be used when running predictions for your intended horizon. Next, we’ll configure specific time-series parameters in an AutoMLConfig object, leveraging Rolling Origin Cross Validation (ROCV) to preserve the integrity of your time-series data.

Delving into the configuration settings, we’ll explore the supported models, including native time-series and deep learning models, and learn how to customize featurization steps for your forecasting task. We’ll also cover optional configurations, such as enabling deep learning, specifying target rolling window aggregation, and handling short or non-stationary time series.

Once your experiment is set up, we’ll walk through running the experiment and retrieving the best model iteration. You’ll then learn how to evaluate the model’s accuracy using a rolling forecast and how to generate predictions into the future, including understanding predictions at specific quantiles of the distribution.

For organizations that require forecasting at scale, we’ll explore the Azure Machine Learning many models solution and hierarchical time-series forecasting, which leverage Automated ML to train and manage millions of models in parallel.

Throughout the article, we’ll reference relevant code samples and example notebooks to provide you with a hands-on learning experience. By the end of this guide, you’ll be equipped with the knowledge and tools to set up Automated ML for your time-series forecasting needs, unlocking the power of Azure Machine Learning to make accurate and informed predictions.