SCA has enhanced its coverage of exponential smoothing methods following the paper, "Exponential smoothing: The state of the art - Part II", by E.S. Gardner Jr. (2006). An extensive set of exponential smoothing methods can now be selected to address different types of trend and seasonal patterns depicted in the two-way table below.
Trend | Seasonality | ||
None | Additive | Multiplicative | |
None | Yes | Yes | Yes |
Additive | Yes | Yes | Yes |
Damped Additive | Yes | Yes | Yes |
Multiplicative | Yes | Yes | Yes |
Damped Multiplicative | Yes | Yes | Yes |
None - Intermittent | Yes | NA | NA |
In addition to exponential smoothing methods for continuous time series, SCA also provides several specialized exponential smoothing methods for intermittent (or interrupted) time series. Such intermittent time series are often encountered in supply-side demand planning applications where demand is lumpy with multiple periods of zero-demand between. Besides the Croston method (1972), which may be the most recognized exponential smoothing method for intermittent time series, variations of the Croston method by Syntetos-Boylan (2001),Teunter and Sani (2009), and others are also available in SCA software.
Apply exponential smoothing in a competely automated fashion, or take full user control over the implementation |
Apply exponential smoothing on an entire spreadsheet of time series data, or on a specific series |
Automatically identify and select the optimal exponential smoothing method based on in-sample and/or holdout-sample forecasting performance, or manually select the exponential smoothing method for a given time series |
Allow the smoothing constants to be estimated within theoretical ranges of the parameters, or further limit the ranges of the individual smoothing constants by setting new upper and lower boundaries |
Choose from several generalized constrained maximization algorithms to determine the optimal smoothing constants, or manually set the smoothing constants |
Automatically adjust for outliers in a time series during parameter optimization and/or forecasting |
Adjust the sensitivity for outlier detection and adjustment, or optionally elect not to employ outlier handling |
Export the assigned model, smoothing constants, and performance measures to a file that can later be imported back into the SCA software and applied to updated time series |
Customize the exponential smoothing procedures using the SCAB34S applet development scripting language. |
TEL: +708-771-4567 EMAIL: sca@scausa.com