Amazon Forecast generates predictions for products with no historical data that is more accurate than before

With Amazon Forecast you can now generate forecasts that are more accurate than before for new products with no history. Forecast is a fully managed service that uses machine learning (ML) algorithms to deliver highly accurate time series forecasts for inventory planning, supply-chain management, and workforce planning - without requiring any machine learning experience.

Cold Start Forecasting is common in industries such as retail and manufacturing, where there is a need to create a forecast for a brand new item, but there is no historical data available. This frequently occurs when a product is introduced to the market for the very first time and has not been offered by the business before.

Traditional time series forecasting methods rely heavily on historical time series of each individual product, making them ineffective at forecasting demand for new products. With this launch, we’re excited to announce a new approach to generating forecasts for products with no historical data. In this new approach, we use item metadata such as color, brand name, or product category that are contained in your dataset to identify similar products and leverage those products to generate a forecast for the product with no history. By inferring based on similar products, we are able to generate forecasts that are more accurate than before.

This feature is available in all Regions where Forecast is publicly available. For more information about Region availability, see AWS Regional Services. To get started on using Amazon Forecast for cold start forecasting, refer to the AWS developer guide or Github notebook.