Scripting language for distributed processing.
A resilient, distributed and optimized generic time series analysis scripting environment supports fast, in-memory time series analysis. The scripting language is optimized and compiled for the machine it's running on – no need to rewrite code for different machines.
Automatic time series analysis & forecasting
TSMODEL procedure includes several function packages that perform a particular task in the time series analysis process. Execute user-defined programs to convert time-stamped transactional data into a time series format, then generate forecast models automatically.
External segmentation of project data
Partition project input data into segments of similar time series based on the nature of the data (e.g., slow moving, new or seasonal items). Then model each segment separately in the project pipelines. This lets you tune modeling strategies to better model patterns or characteristics of the time series in each segment.
Neural networks (NNs) & machine learning
Modeling strategy nodes incorporate NNs through a panel series, multistage (NN/regression + time series), or stacked model (NN + time series) framework. Generate features and train NNs, create a forecasting methodology that combines signals from different model types, and address problems with both time series characteristics and a nonlinear relationship between dependent and independent variables.
Add events to models
Model the effect of events (holidays, retail promotions, natural disasters, etc.) on dependent time series to improve model accuracy. Includes default prebuilt events (e.g., major holidays), and you can add others from an external event repository.
Highly flexible forecast override
A powerful manual override capability lets you make customized adjustments to specific filters or groups of time series defined by attributes, not just by hierarchical variables.
Support for APIs and other programming languages
Includes a broad range of built-in forecasting models and lets you customize models that work well with your data. Use public REST APIs to add SAS Analytics to other applications. Call analytical actions, procedures and APIs from SAS, Python, R, Java and Lua.
Hierarchical reconciliation
Models and forecasts each series in a hierarchy individually, then reconciles top-down at multiple levels. Adjust a forecast at any level, and apportion it to lower levels to keep the hierarchy consistent. Individual forecasts (by products, locations, etc.) roll up to the top number.
Additional forecasting procedures
Includes access to SAS/ETS® and SAS® Forecast Server procedures, enabling you to address virtually any forecasting and time series analysis challenge.
Automatically produce large-scale time series analyses and hierarchical forecasts – without human involvement. Reduced manual intervention means there's less chance of personal bias in the forecasting process. Fewer resources are required, and because forecast analysts don't have to build and monitor forecasting models for every time series, they can focus on more strategic, high-value forecasts or problems that aren't suitable for automation.
Manage your organizational planning challenges by generating forecasts on an enterprise scale – quickly, automatically and as accurately as you can reasonably expect, given the nature of the behavior being forecast. The software delivers results for millions of forecasts at breakthrough speeds, enabling you to plan more efficiently and effectively for the future.
Business drivers, holidays or events that affect the forecasting process are selected automatically from variables supplied to the system in the visual modeling process. You also have the flexibility to manually override forecasts based on groups that are defined using attributes, not just hierarchical variables. The resulting forecasts better reflect the intricacies of the situation.
This solution runs on SAS® Viya®, which has the breadth and depth to conquer any analytics challenge, from experimental to mission critical. SAS Viya extends the SAS Platform to enable everyone – data scientists, business analysts, developers and executives alike – to collaborate and realize innovative results faster.
COMMUNITY
Ask questions, share tips and more in the SAS Forecasting and Econometrics Community.
REPORT
SAS is a Leader in The Forrester Wave™: Multimodal Predictive Analytics And Machine Learning (PAML) Platforms, Q3 2018.
WHITE PAPER
Find out why forecasts are often wrong, and learn how you can improve them.
Check out these products related to SAS Visual Forecasting, built on the powerful SAS® Platform.
SAS/ETS®Model, forecast and simulate business processes using econometric capabilities, time series analysis and time series forecasting.
SAS® Forecast ServerProduce large numbers of forecasts quickly and automatically to improve planning and decision making.
SAS® Visual AnalyticsVisually explore all data, discover new patterns and publish reports to the web and mobile devices.
SAS® Viya®Conquer your analytics challenges, from experimental to mission critical. SAS Viya extends the SAS Platform to enable everyone – data scientists, business analysts, developers and executives alike – to collaborate and realize innovative results faster.
