Forecasting
As of the 2025.10 release, Manage Schedule is available for adoption with Limited Availability.
Forecasting is an updated module developed to facilitate the creation, viewing, and administration of driver forecast. Using advanced machine learning techniques, the forecasting engine delivers accurate and flexible predictions across multiple time granularities, from 15-minute intervals to weekly forecasts. This produces precise demand insights for diverse operational needs.
The enhanced forecasting module offers these capabilities:
- Multi-Granularity Forecasting: Supports forecast horizons from 15-minute intervals to broad weekly overviews, optimizing both short-term tactical and long-term strategic planning.
- Automated Forecasting Pipeline: Provides a user-friendly, end-to-end pipeline where feature engineering, model selection, optimization, and retraining are fully automated, reducing manual effort and ensuring accuracy.
- Scalable Structure: The Forecasting module uses a built-in microservices architecture on AWS and supports real-time scaling to handle large data sets and variable workloads.
- Driver-Linked Forecasting: Forecasted driver values used as key variables influencing demand are integrated with staffing requirements, ensuring accurate scheduling and allocation of resources.
- Advanced Forecasting Model (Multi-Series): Many related time series are forecast together using a shared model. It captures common patterns across series such as stores, items, or departments.
- Robust Forecast Benefits: Enhances accuracy by leveraging adaptive learning, handling sparse or noisy data, generalizing across series, and enabling advanced feature engineering.
- Enhanced User Experience: Offers a modern interface with intuitive navigation, with dynamic charts and tables. This enables stakeholders to interpret complex data and make informed operational decisions.