For the last several decades, autoregressive-integrated moving average (ARIMA) modeling has been a dominant paradigm for statistical analysis of time series data.  In the recent years, however, the analyses based on a larger class of models, the linear state space models (SSMs), have gained popularity.  This shift is mainly spurred by the vastly improved modeling flexibility offered by SSMs, and the ready availability of easy-to-use software for SSM-based modeling.  The SSM and UCM procedures in SAS/ETS® and the CSSM procedure in VIYA/ECONOMETRICS® provide state-of-the-art tools for SSM-based analysis of a wide variety of sequential data types, which include univariate and multivariate time series, panels of univariate and multivariate time series, and data that are generated by multi-level, multi-subject, longitudinal studies.  This presentation provides an overview of the modeling capabilities of these procedures.  Without assuming familiarity with the SSMs, the modeling techniques are introduced using easy-to-follow, real-life examples and useful references are provided for additional information.

Rajesh Selukar is a Principal Statistician Developer in the Advanced Analytics division at SAS.  His primary responsibility is research and development of time series analysis software.  Over the years at SAS, he has developed many computational libraries, SAS procedures, and VIYA actions. His CSSM, SSM, and UCM procedures provide state-of-the-art tools for state space modeling of time series and longitudinal data.  He is also part of the development team of SAS Forecast Studio, a large-scale time series forecasting application.  He has a Ph.D. in Statistics from the University of North Carolina at Chapel Hill.