Use weather data to improve forecasts with Amazon SageMaker Canvas

This article discusses how to use weather data to improve time series forecasting models built with Amazon SageMaker Canvas, a no-code machine learning environment.

Specifically, the article covers:

• Business use cases for time series forecasting and how weather data can improve forecast accuracy
• Finding a suitable weather data provider and evaluating factors like price, data capture method, time resolution, time coverage, geography, and weather features
• Building a weather data ingestion process to harvest, normalize, and combine weather data with business data
• An example workflow using SageMaker Canvas to geocode locations, retrieve weather data from a provider, normalize weather features, and combine with business data
• Using SageMaker Canvas to build time series forecasting models with and without weather data, and evaluating feature importance to determine relevant weather signals
• Conclusion and recommendations for getting started with weather-based forecasting using SageMaker Canvas