Time Series Model Package
TSM Package Summary
Table 1 lists the objects that are contained in the TSM package. These objects are designed to provide access to various univariate time series model families.
Table 1: Objects in the Time Series Model Package
| Object | Description |
|---|---|
| Central Objects | |
| TSM | Time series model object (Note 1) |
| CFC | Combine Forecasts object (Note 2) |
| Time Series Model Specification Objects (Note 3) | |
| ARIMASPEC | Autoregressive integrated moving average model specification object |
| ESMSPEC | Exponential smoothing model specification object |
| EXMSPEC | External model specification object |
| IDMSPEC | Intermittent demand model specification object |
| RNNSPEC | Recurrent neural network (RNN) specification object |
| UCMSPEC | Unobserved component model (UCM) specification object |
| Time Series Model Collector Objects (Note 4) | |
| TSMFOR | Time series model forecast collector object |
| TSMOUTOPT | Time series RNN model optimization history collector object |
| TSMPEST | Time series model parameter estimates collector object |
| TSMRNNSTAT | Time series RNN model fit statistics collector object |
| TSMSPEC | Time series model specification collector object |
| TSMSTAT | Time series model fit statistics collector object |
| Time Series Model Repeater Objects (Note 5) | |
| TSMINEST | Time series model input parameter estimates repeater object |
| TSMINSPEC | Time series model input specification repeater object |
Note:
The TSM object is the central hub that interacts with all other objects. It executes and encapsulates the computational services for all univariate time series models.
The Combine Forecasts (CFC) object allows you to perform forecast combinations.
The time series model specification objects enable you to specify the characteristics of a time series model to be executed by a TSM object.
The time series model collector objects operate on TSM objects to collect and save various results from the time series model execution (fit statistics, forecasts, parameter estimates, model specifications, and so on).
The time series model repeater objects act as conduits to replay collected model specifications and model parameter estimates as input to other TSM objects.
Figure 1 diagrams the relationships among the objects in the TSM package.
Figure 1: TSM Object Data Flow
