SAS® Forecasting for Desktop
Automatic forecasting
- Forecast for large numbers of products, entities or tasks; optimize model parameters.
- Include or add any number of business factors and events in your models to produce the forecast that best depicts your business.
- Specify holdout samples so that forecasting models can be selected based not just on their fit with past data, but also on their likelihood to predict the future.
- Perform automatic hierarchical forecasting by defining the hierarchy and model parameters through an interactive graphical interface.
- Define your own flexible hierarchies to make them more suitable for your forecasting challenges.
- Reconcile up and down the hierarchy, preserving locked forecast values.
- Generate exception reports based on sound statistical logic and business rules.
- Support forecasting as an ongoing and repeated process that fits into your organizational planning workflow.
- Publish results via hard copy, the company portal or the Internet.
Scalability & modeling
- Choose the level of automation for the forecasting process.
- Re-diagnose and identify candidate models, re-estimate existing model parameters or generate forecasts using existing models and parameters.
- Facilitate ongoing and repeatable forecasting as part of your overall planning process.
- Create ensemble models by combining two or more other models.
Unlimited model repository
- Meet all your forecasting needs with an extensible model repository complete with a broad range of built-in models.
- Time series methods include:
- Single exponential smoothing.
- Holt's/Brown's two-parameter exponential smoothing.
- Winters' three-parameter exponential smoothing (additive/multiplicative).
- ARIMA.
- Causal methods include:
- ARIMAX (ARIMA with intervention and causal variables).
- Lagged variables/transfer functions.
- Dynamic multiple regression.
- Unobserved components model (UCM).
- Allows users to create and add their own custom models using the open model repository.
- Users can create and add their own custom models with the open model repository.
Easy-to-use GUI
- Use automated forecasting and set up the hierarchy, parameters and business rules through an interactive graphical interface.
- Employ an eight-step project wizard that walks analysts through the steps to set up a new forecasting project or work with existing projects.
- Easily add events that might influence the forecast – such as holidays and promotions – through an events dialog box.
- Guide analysts through the development of custom forecasting models, built from scratch or based on models in the repository, using a model-builder dialog box.
- Give analysts multiple ways to work with the forecast results using forecast, series and model views.
- Let analysts use the scenario analysis view to create what-if scenarios – such as changes to pricing or promotions – and determine their likely effects on future demand.
- Identify exceptions automatically based on sound statistical logic and business rules.
- Manually adjust forecast values; override the locking facility if needed.
- Publish results automatically via hard copy, company portal or Internet using extensible reporting tools.
Event-modeling console
- Use the interactive GUI to include predefined holiday events (e.g., Thanksgiving, New Year's Day).
- Realign dates automatically for moving holidays (e.g., Easter).
- Use the custom event creator with four event types: pulse, ramp up/down level shift, temporary.
What-if analysis & scenario planning
- Use the Scenario Analyzer interface (which includes what-if planning capabilities) to change model parameter estimates and determine the effects on forecasts.
Data preparation
- Automatically aggregate raw transaction data by time period into time series data.