Amazon Ensemble uses a combination of algorithms that are best suited for your data. Previously, you needed to manually create multiple forecast models and execute multiple forecast actions to find the best outcome for each data collection item. Because of the combined approach, Amazon Ensemble offers greater forecast accuracy as compared to individual algorithms. It provides a combined result for every item in the data collection.
The algorithm provides impact scores that quantify the relative impact of each data collection component (those can be based on related data, attributes, or built-in holidays). In turn, the impact scores determine whether the inputs either increase or decrease of the forecast values.
Amazon Ensemble also:
- Helps you understand which factors, such as sales promotions, production costs, or seasonal trends have the largest influence on your forecast.
- Offers explainability details.