Overview (TCM MODEL command)

The TCM MODEL procedure builds temporal causal models. In temporal causal modeling, you specify a set of target series and a set of candidate inputs to those targets. The procedure then builds an autoregressive time series model for each target and includes only those inputs that have a causal relationship with the target. This approach differs from traditional time series modeling where you must explicitly specify the predictors for a target series. Since temporal causal modeling typically involves building models for multiple related time series, the result is referred to as a model system.

In the context of temporal causal modeling, the term causal refers to Granger causality. A time series X is said to "Granger cause" another time series Y if regressing for Y in terms of past values of both X and Y results in a better model for Y than regressing only on past values of Y.

Data Structures

Temporal causal modeling supports two types of data structures.

Column-based data
For column-based data, each time series field contains the data for a single time series. This structure is the traditional structure of time series data, as used by the Time Series Modeler.
Multidimensional data
For multidimensional data, each time series field contains the data for multiple time series. Separate time series, within a particular field, are then identified by a set of values of categorical fields referred to as dimension fields.

For example, sales data for different regions and brands might be stored in a single sales field, so that the dimensions in this case are region and brand. Each combination of region and brand identifies a particular time series for sales. For example, in the following table, the records that have 'north' for region and 'brandX' for brand define a single time series.

Table 1. Multidimensional data
date region brand sales
01/01/2014 north brandX 82350
01/01/2014 north brandY 86380
01/01/2014 south brandX 91375
01/01/2014 south brandY 70320
01/02/2014 north brandX 83275
01/02/2014 north brandY 85260
01/02/2014 south brandX 94760
01/02/2014 south brandY 69870

Basic Specification

The basic specification is the DATASETTINGS subcommand with the TIMETYPE and TIMEVAR keywords, a FIELDSGROUP subcommand, and a FIELDS subcommand with the TARGETS and INPUTS keywords.
  • For multidimensional data, the DIMENSIONS keyword on the DATASETTINGS subcommand must be specified.
  • If the active dataset has a date specification, then the DATASETTINGS subcommand is not required.
  • If the observations are defined by record order, then the TIMEVAR keyword is not included. Record order means that the first record represents the first observation, the second record represents the second observation, and so on.

Syntax Rules

  • The following subcommands are global and must appear before any FIELDSGROUP subcommand: DATASETTINGS, DATAFILTER, BUILDOPTIONS, TARGETFILTER, SERIESFILTER, FILTEREDOUTPUT, SYSTEMOUTPUT, SAVE, OUTFILE, and PASSPROTECT.
  • Multiple SERIESFILTER subcommands are allowed.
  • Multiple FILTEREDOUTPUT subcommands are allowed. If the same value of the INCLUDE keyword appears on multiple FILTEREDOUTPUT subcommands, then the specifications for only the last such subcommand are honored.
  • Multiple FIELDSGROUP subcommands are allowed, but at least one is required. The specified set of FIELDSGROUP subcommands defines a single temporal causal model system.
  • Multiple FIELDS subcommands can be specified for each FIELDSGROUP subcommand.
  • Global subcommands can be specified in any order.
  • Each keyword can be specified only once within a subcommand.
  • Empty subcommands are not allowed, with the one exception that the FIELDSGROUP subcommand is always empty.
  • Parentheses, equals signs, and slashes that are shown in the syntax chart are required.
  • The command name, subcommand names, and keywords must be spelled in full.

Operations

  • The procedure honors the USE command but it also has settings for specifying the estimation period. Settings on the ESTIMATIONPERIOD keyword are applied after any case selection that is specified with the USE command.
  • The procedure does not honor the PREDICT command. The TCM MODEL procedure provides an option to specify the forecast period.
  • The procedure does not honor the TSET command. The TCM MODEL procedure provides options for handling missing values.
  • The procedure does not honor the following commands: MODEL NAME or SAVE MODEL. The TCM MODEL procedure provides the option to save the model system.
  • The TDISPLAY command does not display model systems that are saved from TCM MODEL.

Limitations

  • The SPLIT FILE setting is ignored by this procedure.
  • Case weighting is ignored by this procedure.